Downloaded from UvA-DARE, the institutional repository of the University of Amsterdam (UvA)

Transcription

1 Downloaded from UvA-DARE, the institutional repository of the University of Amsterdam (UvA) File ID Filename Version uvapub: Thesis unknown SOURCE (OR PART OF THE FOLLOWING SOURCE): Type PhD thesis Title Essays in law and economics Author(s) E.S. Hendriks Faculty FEB: Amsterdam Business School Research Institute (ABS-RI) Year 2014 FULL BIBLIOGRAPHIC DETAILS: Copyright It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content licence (like Creative Commons). UvA-DARE is a service provided by the library of the University of Amsterdam ( (pagedate: )

2 Essays in Law and Economics Eva Susanne Hendriks Eva Susanne Hendriks Eva is both a lawyer and an economist who is interested in the interplay between the two doctrines. Her research includes the analysis of effectiveness and efficiency of public policy, comparative legal studies and the economic analysis of legislation. Eva holds an LL.M. in International and European Law from the Erasmus University in Rotterdam, and an M.Phil. in Economics from the Tinbergen Institute in Amsterdam. She also holds an M.Sc. and a B.Sc. in Economics and an LL.B. in Dutch Law from the Erasmus University in Rotterdam. Essays in Law and Economics This dissertation bundles together a number of essays that find themselves on the intersection of the fields of law and economics. It shows that economics can help determine the rationale of individuals behaviour both in situations of certainty and uncertainty, by looking at incentives and consequences of behavioural choices. Economics can create insights into the rationale and goals of legislation and it can also explain how preferences and information asymmetries influence the outcomes of negotiations between multiple parties. If there is a central thesis to this dissertation, it is that legal issues can benefit from economic analysis in every stage of the legal process. Albeit by explaining legal constructs from a behavioural perspective ex post, or by using economic concepts to inform legislative decisions ex ante. Universiteit van Amsterdam 581

3 ESSAYS IN LAW AND ECONOMICS

4 ISBN Cover design: Crasborn Graphic Designers bno, Valkenburg a.d. Geul This book is no. 581 of the Tinbergen Institute Research Series, established through cooperation between Thela Thesis and the Tinbergen Institute. A list of books which already appeared in the series can be found in the back.

5 ESSAYS IN LAW AND ECONOMICS Academisch proefschrift ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus prof. dr. D.C. van den Boom ten overstaan van een door het college van promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel op vrijdag 6 juni 2014, te 14:00 uur door Eva Susanne Hendriks geboren te Grave

6 Promotie commissie Promotor: Prof. dr. G. Dari Mattiacci Overige Leden: Prof. dr. A.W.A. Boot Dr. U. Endriss Prof. dr. mr. S.A.C.M. Lavrijssen Prof. dr. M.G. Faure Prof. dr. S. Frankel Dr. A. Arcuri Faculteit Economie en Bedrijfskunde

7 Acknowledgements Reaching this milestone in my life makes me reflect on the paths that were taken along the way, and even more so on the people that contributed to making that journey a pleasant and successful one. Although these acknowledgements will inevitably be incomplete, there are a few people who have been essential in my life and accomplishments so far. First of all I am indebted to my promotor: Giuseppe, who has the talent to keep asking questions that push my thinking further and further. I have learned more from you than I could have fathomed at the start of this endeavour. I am deeply grateful to Alessandra Arcuri and Susy Frankel for their advice throughout the writing of this dissertation, and I would also like to thank the other members of my reading committee, Arnoud Boot, Ulle Endriss, Saskia Lavrijssen, and Michael Faure for taking the time to read and assess my work. Most of the work for this dissertation was done at the Amsterdam Centre for Law and Economics (ACLE) at the University of Amsterdam, and the Tinbergen Institute (TI). ACLE provided me not only with a place to work, but also with great company of helpful and inspiring colleagues, many of whom I now consider friends. Sidonie, José, Chandra, who have always lent me a helping hand when I needed one. Jeroen, Mark, Timotej, Martijn and Maarten-Pieter, who have made my time at ACLE both memorable and enjoyable. And last, but certainly not least, Josephine who started out as my roommate in the office, and ended up as my dear friend and neighbour. My time at TI has not only contributed massively to my academic development, but has provided me with many lasting friendships, experiences and life lessons which have contributed to making me the person I am today. Thomy and Lygia: you are always welcoming to me: sharing a dinner, an apartment or baby advice. Christel, the Marcels, Stephen and Amanda, thanks for the many dinners and conversations shared. Thank you Pedro, for the many coffee breaks and chats, and Lerby, for making me laugh, and becoming my neighbour. Before my time in Amsterdam, my academic career started in Rotterdam. My time there has been made especially memorable by numerous friends. Stéphanie, Esther, Tamara, thanks for the many lovely Monday night dinners, drinks, dances and for making the days spent in the library enjoyable. Thanks Carline, Claudia, Rieke and Sandra, for sharing many dinners, teas, copies of notes, and general good times. A special word of

8 gratitude to Lara, together we were the slowest rowing team in the history of Skadi, but our conversations, and laughter, and your support in times both good and bad spans oceans. A special thanks to Laura, Mimi and Charlotte, for our endless Christmas dinners, held at any month in the year. Thanks a lot to Erika, for our chocolate cake dates in Geneva, and thank you Nieves, for always peer-pressuring me into doing exercise with you. My rock of support has always been my family, who deserves my gratitude more than anyone else. I would not have been where I am today without my parents, who have always remained understanding and undyingly supportive when I decided to pursue yet another academic degree, or take a new direction in my life. I am sorry for the sleepless nights I have caused you, but I am proud that I can at least uphold the family tradition of not graduating without a child. My brother Thijs, whom I think is the only person on the planet to understand and share my sense of humour. Emmy, for making me happy by how much you mean to my brother. Janneke, the strongest person I know, who is a role model to me for what it means to be a good person. Gijs, for bringing me the two people that are so important in my life. Elisa and Madelief, who have taught me what it means to love unconditionally. Last but not least I owe gratitude to Ben, for your patience with me and for everything, past, present and future. Thank you for sharing your life, your country and your dreams with me. And finally there is my baby, who at the moment of writing is yet to be born. I already look forward to rediscovering the world through your eyes.

9 Contents 1. Introduction Relative Fault and Efficient Negligence: Comparative Negligence Explained Introduction Theories of sharing To share or not to share How to share An illustration of the results Formal analysis Setup and efficiency equivalence Equilibrium deviations from due care are efficient Comparative negligence stimulates efficient negligence Apportioning damages according to relative fault A numerical example Extensions Sequential care choices Courts err with respect to the parties' costs of care Multiple tortfeasors Third-party externalities Litigation rate Conclusion A Comparative Review of the Protection of Traditional Knowledge a Theory of Three Approaches Introduction Legislation on the protection of TK description of the database Selection of the data Categorization... 49

10 3.3. Analysis Reported legislation Regional preferences The three approaches to the protection of TK How much of the legislation is explained by the theory of the three approaches? Factor analysis Conclusion, discussion and further research The Way Forward for International Negotiations on the Protection of Traditional Knowledge Introduction Methodology Current and past international negotiations Integral international agreements Global negotiations on the protection of specific forms of TK Regional negotiations on integral TK protection Aspects influencing the creation and working of international instruments An international framework of protection - Potential outcomes of negotiations Outcomes of negotiations on an integral framework to protect all forms of TK Potential outcomes of negotiations on approach-specific agreements Aspects influencing the efficiency and effectiveness of international agreements Economies of scale Challenges for a one-size-fits-all framework Cross-border aspects

11 Learning effect Economies of scope Diseconomies of scale and scope, or the feasibility of reaching an international agreement Identification of opportunities and challenges for international agreements on TK conclusion A Generalized Jury Theorem Introduction Model: information-aggregation as sampling The correct decision (common goal) Jurors and signals Unbiasedness and accuracy of the jurors The jury decision Framework The Generalized Jury Theorem Applications Binary decisions and decisions over a continuum Optimal use of available information Conclusions Conclusions and future research Appendices Appendix 2.1 Solution of the discrete game Appendix 2.2 Proof of Lemma Appendix 2.3 Proof of Proposition Appendix 2.4 Proof of Corollary Appendix 2.5 Proof of Proposition Appendix 2.6 Proof of Proposition Appendix 3.1 Definitions as used by WIPO

12 Appendix Genetic Resources Appendix Traditional Knowledge Appendix Traditional Expressions of folklore Appendix 3.2 Correlation coefficients between the region and the reported legislation on TK and Biological TK Appendix.3.3 Correlation coefficients between the regions and the content of the reported legislation on TK protection Appendix 3.4 Correlation coefficients within the economic empowerment approach Appendix 3.5 Correlation coefficients within the preservative protection approach Appendix 3.6 Correlation coefficients within the cultural integrity approach Appendix 3.7 Factor analysis Appendix Analysis: does a factor analysis support the theory of the three approaches? Appendix How much of the legislative choices is explained by the theory of the three approaches? Appendix 5.1 Proof of the Generalized Jury Theorem Appendix 5.2 A jury decision modelled by means of a Bernoulli (p) distribution is regular Appendix 5.3 A Jury decision modelled by means of a distribution is regular Appendix 5.4 Optimal weights with heterogeneous jurors References Bibliography Domestic legislation and regulations Documents of international organizations Case law Samenvatting (Summary in Dutch)

13 Tinbergen Institute Research Series

14

15 List of figures Figure 1. Apportionment of damages under negligence rules Figure 2. The parties expected costs under negligence rules Figure 3. The parties expected costs under negligence rules Figure 4. Relative number of states per region that have reported legislation Figure 5. Three approaches for legislation on TK-protection Figure 6. Principal factor analysis Figure 7. The link between factor scorings and approach taken Figure 8. Three approaches as supported by the factor analysis Figure 9. Factor scorings for countries for factor 1 and Figure 10. Factor scorings for countries for factor 1 and Figure 11. Issues of TK protection Figure 12. Potential for efficiency gains from economies of scale Figure 13. Expected loss of effectiveness from the lack of a one-size-fits-all framework in an international framework of protection Figure 14. Advantage of an international framework in effectiveness to counter the negative effects of cross-border aspects Figure 15. Potential loss of effectiveness in an international framework due to the loss of a learning effect Figure 16. Potential for efficiency due to economies of scope Figure 17. Feasibility of the creation of an international framework Figure 18. Information-aggregation and estimation theory Figure 19. Direct versus indirect majority Figure 20. Correlation coefficients between region and reported Figure 21. Correlation coefficients between region and content of the reported legislation Figure 22. Correlation coefficients within the economic empowerment approach Figure 23. Correlation coefficients within the preservative protection approach Figure 24. Correlation coefficients within the cultural integrity approach Figure 25. Principal factor analysis regions included Figure 26. Factor loadings per factor regions included Figure 27. Link between factor scorings and approach taken Figure 28. Link between factor scorings, region and wealth

16 Figure 29. Principal factor analysis Figure 30. Principal factor analysis Figure 31. Link between factor scorings and approach taken Figure 32. Three approaches as supported by the factor analysis Figure 33. Factor scorings for countries for factor 1 and Figure 34. Factor scorings for countries for factor 1 and

17 1. Introduction If there is a central thesis to this dissertation, it is that legal issues can benefit from economic analysis in every stage of the legal process. Whether it is by explaining legal constructs from a behavioural perspective ex post, or by using economic concepts to inform legislative decisions ex ante. This dissertation bundles a number of papers that find themselves on the intersection of the fields of law and economics. Researchers in either field can learn from the insights of the other. Due to the interdisciplinary nature of this dissertation the various chapters are written for varying audiences. Some are directed at legal scholars, introducing insights and techniques from economics, while others are directed at economic scholars, introducing new applications and contexts for well-known concepts. This variation in target audiences is why some chapters display more emphasis on legal issues, and other more emphasis on economic issues. Economics can help determine the rationale of individuals behaviour both in situations of certainty and uncertainty, by looking at incentives and possible consequences of behavioural choices, which is done both in Chapter 2 and in Chapter 5. Economics can also help to create insights into the rationale and goals of legislation, which is done in Chapters 2 and 3, where economic methods are applied to legal questions. Economics can also explain how preferences and information asymmetries influence the outcomes of negotiations between multiple parties, which is done in Chapters 4 and 5, where economic concepts are applied in legal contexts. Chapter 2 discusses the rationale of negligence rules. It deals with situations where individuals make decisions on whether to take a costly measure to prevent a possible harm from occurring. For example a cyclist can decide whether to wear a helmet to protect himself from head injury. However, whether or not he might get a head injury also depends on the behaviour of other individuals, in this example a driver. The driver can decide whether to look in his mirrors before he turns a corner and with that influence whether or not the cyclist might get a head injury. If the harm occurs then there is also the judge who has to determine ex post whether or not the cyclist did wear a helmet and whether the driver did look over his shoulder and whether we should expect either individual to do so to determine whether either of the individuals was at fault. And then lastly, the judge has to apply the negligence rule to determine which individual has to bear the damages. 1

18 There are various negligence rules that deal differently with determining who has to pay the damages in a situation where there are two individuals involved and neither took measures to prevent the harm from occurring. However, comparative negligence, which shares the damages between the two individuals, is the most common rule applied both in Europe and the US. 1 In Chapter 2 we use game theory to show that this particular negligence rule creates incentives that induce efficient behaviour in certain situations where other negligence rules do not. A judge can make a mistake in assessing what the optimal standards of care were, for example by setting the due care standards too high, or in the example of the cyclist and the driver, by expecting the cyclist to wear full body armour, and to expect the driver to stop at every intersection to get out of the car and check in every direction whether a cyclist approaches. In such situations, where judges err in assessing the due care levels, negligence can be the optimal behaviour. The intuition behind this is that comparative negligence shares the damages between two individuals involved, as opposed to other negligence rules that appoint the damages to one individual only. When one individual will bear the full cost of the damages the threshold to make negligence a rational choice for that individual is relatively high. And once one individual decides not to be negligent this transfers the full expected cost of damages onto the other individual, if he were to be negligent. This transfer of the expected costs creates a similarly high threshold for negligence for that individual. The sharing of the damages in comparative negligence lowers the expected costs of negligence for either individual, and hence lowers the threshold to make negligence the optimal behaviour for either individual. We show that comparative negligence induces negligence when it is efficient, in situations where other negligence rules do not, and we also show that the optimal sharing rule is one of relative fault, sharing the damages proportional to the individuals relative departures from due care. Chapter 3 shows how the application of statistical methods can shed light in the field of comparative law. A structured overview is created of all existing legislation on a specific topic: the protection of traditional knowledge (TK). Factor analysis allows for an in-depth analysis of underlying concepts of the legislation, that so far were not found through conventional comparative law methods. Through the application of this methodology a theory of three approaches is introduced, explaining that the existing legislation on the 1 Curran (1992); Calabresi (1997 p. 2206); Best (2007); Robinette and Sherland (2003); van Dam (2006, pp ); and Artigot I Golobardes and Gomez Pomar (2009, pp 48-52). 2

19 protection of TK can be subdivided into three distinct approaches, each with specific characteristics such as goals of the legislation, what it protects TK from, and whom it is protected for. The three approaches found are: the economic empowerment approach, through which the legislator seeks to create opportunities for the economically weaker indigenous groups in society to monetize on their traditional knowledge; the preservative protection approach, in which the legislator seeks to preserve folklore for the future and for the benefit of the country as a whole; and the cultural integrity approach, used to prevent the offensive and inappropriate use of sacred culture, historical objects and authentic new traditional knowledge products. The Theory of the Three Approaches is shown to explain up to 81% of the variation found in the legislation. Chapter 4 applies economic concepts, mainly borrowed from economics of federalism, to analyse the potential benefits from hypothetical international agreements. Chapter 4 continues in the realm of TK, it applies the results of Chapter 3 and uses the Theory of the Three Approaches as a proxy for the preferences of negotiating parties in international negotiations on the protection of TK. It analyses reasons why international negotiations on this topic have been unsuccessful thus far. It discusses whether the possibilities exist for gains in efficiency and effectiveness in the regulation of the protection of TK if international agreements are formed. It also discusses where potential losses in efficiency and effectiveness might occur if international agreements are formed. Combining these results leads to a conclusion in which approaches there is potential gain from international negotiations, and in which approaches there is not. Chapter 5 analyses the rationale of individuals behaviour in situations of uncertainty. It looks at information aggregation in groups of individuals, who are trying to make an optimal decision under uncertainty, based on the information present among the individual group members. We use statistics to develop a framework of analysis capable of dealing with various forms of choices: the Generalized Jury Theorem. We show that the Generalized Jury theorem should be used to model the behaviour of jurors, in situations where jurors have a common goal and can communicate before voting. The intuition behind the Generalized Jury Theorem builds on the original premise of Ramond Llull ( ) that the quality of a decision based on the aggregation of individual bits of information is determined by the quality of the individual bits of information and by the way the information is aggregated. The Generalized Jury Theorem shows that as long as the individual jurors share a common goal and are given the possibility to communicate prior to voting, it is in all the jurors interests to truthfully 3

20 reveal their private information. The reason for this is that this will allow them to make a decision that best fits their shared preferences. Once all information is shared, the decision should be reached unanimously as the optimal decision is then known. This means that there is full revelation in equilibrium, irrespective of the voting rule used. Our framework builds on Condorcet s Jury Theorem, however, Condorcet s Jury theorem can only be applied to binary choices, while the Generalized Jury Theorem can be applied to a broader set of choices including to decisions over a continuum. In other words, Condorcet s Jury Theorem can only be applied to juries dealing with yes/no questions, like whether a defendant is guilty, while the Generalized Jury Theorem can be applied to broader questions, like decisions on what the speed limit should be on a motorway. Condorcet s Jury theorem can be seen as a special case of the Generalized Jury Theorem. Chapter 6 provides conclusions and a brief outlook for future research building on the work presented here. 4

21 2. Relative Fault and Efficient Negligence: Comparative Negligence Explained This paper is co-authored with G. Dari-Mattiacci and published in the Review of Law and Economics, June 2013; 9(1): pp ABSTRACT This paper shows that the rule of comparative negligence with relative fault - a sharing of the loss proportional to the parties' relative departures from due care - induces the parties to an accident to be efficiently negligent. Comparative negligence is more efficient than simple or contributory negligence regimes because it serves as a buffer against excessive due-care standards. If due-care standards are too high, comparative negligence facilitates efficient negligence, inducing parties to violate excessive due-care standards only when this is socially desirable. If due-care standards are too low, all negligence rules perform in the same way. Of all possible comparative negligence rules, relative fault provides for the sharing rule that maximizes this effect. The same principle also applies to the contribution rule among multiple tortfeasors. 5

22 2.1. Introduction The efficiency of comparative negligence poses a persistent puzzle to law and economics scholarship. Although many notable attempts have been made to pinpoint the reasons for its widespread use, a generally accepted theory of comparative negligence has eluded scholarly efforts. Moreover, the actual apportionment of damages between two negligent parties is a question that rarely emerges in digested opinions and has not been resolved by scholarship. 2. In Wilson v. R&C Serv. Co. the jury determined that the plaintiff was 29% comparatively negligent. 3 Although juries do not shy away from very precise determinations of comparative fault, it is not clear how such numbers are or ought to be produced. When an accident can be prevented if both the victim and the injurer take care and the costs of care are $1 for the injurer and $99 for the victim, which party should bear the largest share of damages? Intuition and precedent 4 suggest that the party that could have taken care at a negligible cost and failed to do so bears greater responsibility for the accident and should bear a correspondingly greater share of damages. We will show that this argument is misleading and the party with the greater cost of precaution should bear more liability. We propose a novel theory that explains why comparative negligence is more efficient than its all-or-nothing alternatives (simple and contributory negligence) and makes an efficiency case for a particular sharing of the loss between negligent parties proportionally to relative fault that is, the parties relative departures from due care. Consider the following example: 2 The apportionment of damages is generally determined by the trier of fact in a non-transparent way. The Uniform Comparative Fault Act, Sec. 2.b states that In determining the percentages of fault, the trier of fact shall consider both the nature of the conduct of each party at fault and the extent of the causal relation between the conduct and the damages claimed. See also Watson v. State Farm Fire & Cas. Ins. Co., 469 So. 2d 967 (La. 1985) and Restatement (third) of Torts: Apportionment of Liability 8 (2000) detailing the factors that should bear on the determination of the parties' comparative fault. Cfr. (Parisi 2004) dealing with the sharing of the loss between two non-negligent parties. 3 CV S, 2003 WL (Conn. Super. Ct. Feb. 19, 2003). Different numbers emerge from litigated cases: the plaintiff's comparative fault was assessed at 20%; in Allen v. Perry, 722 S.W.2d 98, 100 (Mo. Ct. App. 1986), at 33%; in Griffin v. LeCompte, 471 So. 2d 1382, 1389 (La. 1985), at 40%; in Jensen v. ARA Services, Inc., 719 S.W.2d 121, 122 (Mo. Ct. App. 1986), at 60%; in Vincent v. Pabst Brewing Co., 47 Wis. 2d 120, 123, 177 N.W.2d 513, 514 (1970), and at 66%; in Griffin v. LeCompte, 471 So. 2d 1382, 1389 (La. 1985). 4 Dobson v. Louisiana Power and Light Co., 567 So.2d 569 (La. 1990). 6

23 Example 1. An accident can be prevented if both Xavier (the injurer) and Yvonne (the victim) spend $60 on care. Care taken by only one party has no effect. 5 If an accident occurs, damages amount to $100 Clearly, the accident should not be prevented because the total costs of care ($60+$60=$120) are larger than the accident loss ($100). A court should not find the injurer negligent in this case. To apply the Hand formula 6 correctly, the court needs to estimate the parties' costs of care, compare these with the damages resulting from the accident, and find that the former are larger than the latter. What if the court makes a mistake and holds the injurer liable? Under simple negligence, the injurer faces a choice between spending $60 on care and paying $100 in damages and will opt for the former. Given that the injurer can be expected to take care, the victim faces a choice between preventing the accident ($60) and bearing the accident loss ($100), and will also take care. Under simple negligence, a wrong determination of negligence induces both parties to take precaution in a case when they should not. (The same result occurs under contributory negligence.) Under comparative negligence, if both parties are found negligent the loss is shared (say, at 50%). If a party believes that the other party will not take care, the best choice is not to take care because the per capita liability share amounts to $50 while taking care would cost $60. Thus, comparative negligence leads to an equilibrium in which both parties are efficiently negligent. 7 Unlike all-or-nothing rules, the sharing of the loss induces the parties to choose the same action that they would choose if they could cooperate and allows them to offset the error made by the court. This equilibrium is preferable for the parties and is the socially desirable outcome. 8 The example above shows that comparative negligence could create a sharing of the loss that reduces social costs compared to all-ornothing rules. Since the optimal sharing will not necessarily be 50%, what sharing rule should the court apply when apportioning damages between the parties? 5 The assumption that unilateral care does not affect the accident loss is innocuous and is made only to simplify the examples. The model that we present in section 2.4 considers the general case where unilateral care may or may not reduce the accident loss. 6 United States v. Carroll Towing Co., 159 F.2d 169 (2d. Cir. 1947). 7 The notion of efficient negligence has been introduced by Grady (1998) in the context of inadvertent violations of the due-care standard. See also Grady (1990). In the conclusion, we will explain how our theory relates to Grady s theory. 8 The equilibrium where both parties take care persists but is unlikely to be chosen because it entails higher costs for each party. Mixed strategy equilibria are unstable. In contrast, the negligence equilibrium is stable and is the Pareto efficient equilibrium (and hence a focal point). 7

24 Example 2. An accident can be prevented if both Xavier (the injurer) and Yvonne (the victim) take care. Care taken by only one party has no effect. Care costs Xavier $30 and Yvonne $90. If an accident occurs, damages amount to $100. In Dobson v. Louisiana Power and Light Company, 9 the court concluded that if the defendant's cost of care is less than that of the plaintiff, then the defendant's fault is the greater of the two. Thus, Xavier should bear a greater share of liability than Yvonne. The court does not offer a precise formula to calculate such shares but any apportionment rule that assigns more than 50% of the damages to Xavier would induce both parties to take care: Xavier would prefer to take care ($30) rather than bear his share of damages ($50 or more). Given this choice, Yvonne would prefer to spend $90 on care and avoid the accident rather than bear the accident loss ($100). Although appealing and seemingly fair, a sharing rule that assigns more liability to the party with the lower cost of care fails to provide the parties with incentives to be efficiently negligent. Instead, an optimal sharing rule should apportion damages so that each party bears a share of liability that is less than his or her cost of care in all those cases in which the court implements excessive care standards. Thus, Xavier's share of damages should be less than Yvonne's. A sharing rule that apportions damages proportionally to relative fault - that is, the parties' relative departures from due care 10 - has the very important property of always implementing the optimal sharing. In the context of Example 2, Xavier's fault amounts to $30 because the court maintains that he should have taken care but he did not. Similarly, Yvonne's fault amounts to $90. Therefore, according to his relative fault, Xavier should pay of the damages, while Yvonne should bear the residual. 11 With this apportionment, both parties have an incentive to be efficiently negligent: for Xavier being negligent and paying $25 is better than taking care at a cost of $30 and for Yvonne $75 is better than $90. This behaviour is also the socially desirable course of So.2d 569 (La. 1990). 10 Note that the notion of relative fault could be given a different interpretation: one could argue that the party with the lower costs of care is more at fault than the other party. We have discussed this case in footnotes 4 and 8 and accompanying text. One of the points we make in this paper is that this interpretation is misleading and that relative fault should be defined according to the parties relative departures from due care. 11 Cf. Barnes and Baeverstad (1982, p.284) using this apportionment rule to make a different point. 8

25 action. In our analysis below, we will demonstrate that this error-correction property of relative fault is not a feature of the example chosen but applies generally. Moreover, we will show that relative fault is the only sharing rule that generates incentives for the parties not to take due care whenever due care is socially undesirable and to take due care otherwise. What happens if the court sets due care too low, holding a party non-negligent when in fact he or she should be held negligent? In this case, all negligence rules induce the same behaviour because they only differ with respect to the allocation of the accident loss between negligent parties. The effect of standards of care that are too lax is that parties will be found non-negligent too often. This effect falls outside the region in which the differences among rules matter (which is only when parties are found negligent). If due care standards are too low, all rules perform in the same way and hence our results are not reversed. 12 Therefore, our analysis applies generally to erroneous due-care standards and not only to excessive ones. (A similar argument explains why our findings also apply to the case when due care is too high for one party and too low for the other.) However, note that court assessments of the probability of accidents are made with hindsight and are more likely to be biased upwards rather than downwards. 13 Why should courts make errors with respect to the level of due care if they can correctly assess relative fault? Example 3. The probability that an accident occurs can be reduced from 10% to zero if both Xavier (the injurer) and Yvonne (the victim) take care. Care taken by only one party has no effect. Care costs Xavier $30 and Yvonne $90. If an accident occurs, damages amount to $1,000. Like in the previous examples, the accident should not be prevented because the total costs of care $30+$90=$120 are larger than the reduction in the expected accident loss 10% x $1,000=$100. Assume that the parties did not take care and an accident happened. 12 Consider the following example: An accident can be prevented if both Xavier (the injurer) and Yvonne (the victim) spend $40 on care. Care taken by only one party has no effect. If an accident occurs, damages amount to $100. This is clearly an accident that should be prevented. However, if an accident occurs and the court erroneously concludes that the injurer's care cost was not justified (sets the due care level too low) and thereby finds the injurer non-negligent, parties will not have an incentive to take care. The injurer prefers spending nothing rather than taking care at the cost of $40. In turn, since the injurer can be expected not to take care and the accident cannot be prevented unilaterally by the victim, she has no reason to take care. Note that this result does not depend on how the loss is shared if both parties act negligently. Therefore it is irrelevant which negligence rule is implemented. 13 See Kamin and Rachlinski (1995) showing that, if people are asked to assess the probability of accidents after the fact, their estimates are lager if the accident has actually occurred. 9

26 The court has two tasks: assessing relative fault and setting the due-care standards. Assessing relative fault is relatively straightforward. The court needs to compare the cost of the parties' taken precautions (zero, in the context of the example) with the costs of taking due care. It should be stressed that relative fault does not need to be based on the socially-optimal due-care standards in order to perform its error-correction function. On the contrary, relative fault corrects court errors precisely because it is based on the court's (possibly erroneous) determination of due care. 14 Therefore, it is immaterial which levels of due care the court chooses. In contrast, setting due care at the optimal level requires far more information. In addition to assessing the parties' costs of care, damages, and whether or not the parties took care - which are complex tasks - the court needs to estimate the ex ante probability of accidents. This might prove to be a formidable endeavour, because the probability of accidents is a function of the parties' behaviour. The court needs to calculate not only the ex ante probability of accidents given the parties' taken precautions (10%, in the example) but also the probability of accidents under any alternative behaviour. In order to do so, the court has to establish with some reliability a series of counterfactual claims about what would have happened had the parties behaved differently. In the context of Example 3, it has to be demonstrated that, if the injurer alone or the victim alone had taken care, the probability of accidents would have been 10% and that, if both parties had taken care, the probability of accidents would have been zero. Moreover, in most cases the magnitude of the damage is also a function of the parties' care and calculating damages that would have occurred under hypothetical alternative conditions is an exercise that falls into the realm of educated guesses, at best. Therefore, a court is likely to make errors in the calculus of the optimal due-care standards but is nevertheless in the position of applying relative fault correctly, because the latter task requires much less information than the former. Regulatory agencies or governmental bodies involved in the setting of due-care standards typically face the same difficulties as courts, because they react to information about accidents that have already occurred and attempt to come to an assessment of possible alternative courses of events. The parties often have superior knowledge about the effects of precaution on the probability and magnitude of losses because they were there at the time of the accident Note that in fact in this case the optimal due-care standards are zero for both parties. 15 See Shavell (1984, p. 359), explaining that when parties have superior knowledge about factors such as the benefit of the activity, the cost of reducing risk and the probability and magnitude of losses, liability 10

27 Even if this information is observable ex ante, it is often difficult to prove in court ex post. Hence, courts might incur in errors concerning the optimal due-care standards even when the parties do not make such errors. The case of Li v. Yellow Cab Co. 16 provides an example of a typical traffic accident context in which parties might have more information than courts on the optimal due care standards. When this is the case, comparative negligence implies the indirect use of information that would otherwise be lost. What if parties and courts make the same errors? In this case, all liability rules perform in the same way since the parties act as if the due-care standards set by the court were the efficient ones. 17 Our analysis emphasizes the advantages of comparative negligence when parties have more accurate information than courts and regulators (which may set due-care standards too high or too low) and remains valid when this is not the case (and hence all negligence rule perform in the same way). In Section 2.2, we refer to existant literature on court errors and comparative negligence and explain why our approach differs from previous attempts to solve the puzzle of comparative negligence. In Section 2.3, we present a simple discrete model, in which parties can only choose whether they take care or not, but the costs of care are fixed. The discrete model illustrates the strategy and the results presented in a more general setting in Section 2.4. In Section 2.4.1, we present the setup of a general model with choices of care over a continuum. We analyse the behaviour of injurers and victims under erroneous standards and prove three results. We first show, in Section 2.4.2, that negligence in equilibrium is efficient. Any deviation from the due-care standards that emerges in equilibrium under any rule is efficient from the social welfare point of view because it yields less social costs of accidents than would occur if the parties abided by due care. Second, in Section we show that comparative negligence is the rule that most frequently induces efficient negligence. Finally, in Section 2.4.4, we show that the should be preferred over regulation; otherwise, safety regulation is a better way of alleviating risks. Therefore, the context in which parties are better informed than courts and regulators coincides with the case in which liability is used, while the alternative scenario occurs when regulation in used and hence is not directly relevant for our analysis P.2d 1226 (Cal. 1975) See Grady (1998 p. 416) for a discussion of this case in the context of comparative negligence. 17 If instead courts have superior information, the due-care standard should not be based on information that was not available to the parties ex ante, because the parties will not be able to predict the court decision. Thus, superior information by the court will in most cases not be used in trial (see Dari-Mattiacci and Garoupa (2009) and references therein). Hence, this case reduces to the case in which parties and courts have the same information. If instead regulators have superior information, then regulation of safety should be preferred over liability (Shavell 1984). Hence this case falls outside the scope of this paper, which focuses on tort liability. 11

28 commonly used rule that shares damages according to the parties' respective negligence always induces noncompliance whenever such an equilibrium can exist. In Section 2.5, we extend the analysis in several directions, including sequential care choices, errors in the assessment of the parties' costs of care, multiple tortfeasors, third-party externalities, and effects on the litigation rate. In Section 2.6, we conclude and emphasize the implications of our results Theories of sharing Comparative negligence is both the most common rule in the US and in Europe 18 and a rule that law and economics scholarship has had enormous difficulties justifying. Early studies suggested that comparative negligence, by splitting the loss between two negligent parties, could dilute incentives to take care as compared to simple and contributory negligence. 19 Later, it was shown that, under assumptions of complete information, all negligence rules produced the same (efficient) incentives to take care. Thus, the splitting of the loss between two negligent parties has no effect on the social cost of accidents. 20. The allocative efficiency of all negligence rules poses the challenge to explain why comparative negligence is the most common rule. In fact, prima facie, comparative negligence seems to have a serious disadvantage compared to simpler all-ornothing solutions: by introducing an additional layer of complexity into the trial, comparative negligence could increase the costs of litigation. 21 Scholars have proposed several theories of comparative negligence. 22 These will be discussed below To share or not to share We first review theories that provide bases for some sharing of the loss between negligent parties and then move on to the question of how this sharing should be calculated in practice. 18 Curran (1992); Calabresi (1997 p. 2206); Best (2007); Robinette and Sherland (2003); van Dam (2006, pp ); Artigot I Golobardes and Gomez Pomar (2009, pp Brown (1973); Posner (1977); cf. Posner (2010, p. 222) for a different view. 20 Haddock and Curran (1995) are usually credited for what is known as the allocative equivalence theorem or the efficiency equivalence theorem ; however, an earlier proof of the allocative equivalence of negligence rules can be found in Landes and Posner (1980, p. 539, fn. 51). See Jain and Singh (2002) for a general characterization of liability rules and Jain (2009) for an analysis of incremental liability rules. 21 Posner (2010, p. 222); cf de Mot (2013) showing tht comparative negligence might result in lower litigation costs. We will examine litigation costs in section See Bar-Gill and Ben-Shahar (2003) and Artigot I Golobardes and Gomez Pomar (2009) for two excellent surveys of the literature. 12

29 Fairness If both parties are at fault, there is no reason why one of them should bear the full accident loss. Comparative negligence, by splitting the loss between two parties who are equally to blame for the accident, provides a more equitable solution than simple or contributory negligence Insurance Economic models of accidents are usually grounded in the simplifying assumption that there is a victim (who suffers a loss) and an injurer (who suffers no loss). In reality, however both parties could suffer some loss and the distribution of the harm between the parties may be a matter of chance. Consider, for instance, a collision at sea. A rule that completely bars recovery places a heavier burden on the vessel that suffers more damages; in contrast, a rule of comparative negligence that splits the total loss between both parties guarantees a more predictable and less risky outcome and therefore insures parties against (some of) the risks deriving from accidents (Posner, 2010, p. 223) Lapses into negligence In the real world parties may fail to play equilibrium strategies and it may occasionally occur that a party either lapses into negligence 24 or for other reasons chooses not to take care even when the game-theoretic model of accidents predicts that both parties should take care. It has been argued that in these situations comparative negligence could save the day by creating some incentives to take care to a party that observes or expects the other to behave negligently. 25. Although appealing, the correction-of-lapses theory explains too much. If the victim is expected to lapse into negligence the strongest incentives for the injurer are provided by simple negligence, not by comparative negligence. Similarly, if the injurer can be expected to lapse into negligence, the strongest incentives for the victim are provided by contributory negligence. If both parties can be expected to lapse into negligence, it is not clear whether a rule that produces some incentives for both parties 23 See Cooter and Ulen (1986), Bar-Gill and Ben-Shahar (2003) and Dari-Mattiacci and de Geest (2005) for references. 24 See Grady (1990) and further Grady (1998) on lapses and their relation to comparative negligence, see further section See Orr (1991) and, for a critique, Chung (1993), p. 404 there is no efficiency motivation for favouring one rule over the other. 13

30 (comparative negligence) is to be preferred to rules that produce full incentives for one party only (simple and contributory negligence), or vice versa Least-cost avoidance Orr (1991) claimed that a dilution of incentives can be beneficial if accidents are of the alternative-care type: one party's care is enough to prevent the accident. Then, a properly engineered comparative negligence rule can induce only one party to take care (which is efficient) while contributory negligence could induce both parties to take care. Chung (1993) later showed that this result derived from an implicit assumption concerning the erroneous determination of the negligence standard. If one party's care is enough to prevent the accident, then the negligence standard should concern that party only, while the other party should be considered non-negligent by definition, irrespective of whether he or she took care. Therefore, efficiency can be realized by imposing liability on one party only rather than imposing liability on both parties and then splitting the loss in such a way that incentives for one of them are too low to take care Filtering-out the most harmful accidents More recently, and after the publication of Bar-Gill and Ben-Shahar (2003), a critical appraisal of extant theories of comparative negligence, Dari-Mattiacci and de Geest (2005) have tackled this puzzle anew. They propose a law-enforcement model in which precaution by one party is enough to avoid the accident, choices are binary (violate or comply) and the enforcer is incompletely informed about the parties' precaution costs. The analysis shows that comparative negligence has a filtering effect: it prevents accidents that yield high social costs, but lets accidents that yield low social costs occur. Rules that burden one party, such as simple or contributory negligence, also prevent some accidents while letting other accidents occur, but they do so at random, without any selection. Therefore, comparative negligence improves social welfare by filtering out the most harmful accidents. This analysis, however, is limited to a discrete model of unilateral (only one party can take care at reasonable costs) or alternative care (both parties can take care at reasonable cost but one of them is the least-cost avoider). 26 The theory that we present in the following relates to lapses in a specific way. We do not address the strategic problem of corrective sequential care discussed above. Rather lapses are relevant to out theory insofar as the presence of lapses is known to the parties but is difficult to verify in court; that is, parties are aware of the possibility of lapses but the court is not. If this is the case, courts might make errors in setting the due-care standards while parties know the socially optimal levels of care. To the extent that lapses can be rationalized as a determinant of court errors our theory applies to lapses. 14

31 Evidentiary uncertainty Cooter and Ulen (1986) analysed the functioning of tort liability under evidentiary uncertainty, when parties cannot accurately predict whether the court will find them negligent. Evidentiary uncertainty induces parties to alter their levels of precaution. Cooter and Ulen found that, when parties are symmetrical in their abilities to take precautions, comparative negligence is to be preferred because it generates moderate distortions by both parties, while simple and contributory negligence cause large distortions for one party and small distortions for the other. 27 Bar-Gill and Ben-Shahar (2003) emphasize that this result relies on the hypothesis that moderate distortions are preferable to large ones, which needs not be true in general. In addition, they show that moderate distortions could also be obtained by all-or-nothing rules Self-selection Rubinfeld (1987) proposed a setting in which there is heterogeneity among parties with respect to their costs of care, which cannot be directly observed in court. He finds it optimal for the court to set very high standards - in order to induce parties to be negligent - and employ comparative negligence - in order to induce parties to self-select with respect to their costs of care. The mechanism works because the parties can make small adjustments in their share of liability by taking more or less care. In contrast, under simple or contributory negligence, liability jumps discontinuously from zero to full liability, and adjustments in care are not fine-tuned. Bar-Gill and Ben-Shahar (2003) build on Rubinfeld's analysis and show that information revelation mechanisms can be constructed in response to a broader set of problems and that they can be implemented in all negligence rules, not only under comparative negligence. For instance, if the victim has more information about the harm than the injurer or the court, then a rule that sets damages equal to the expected harm allows victims to self-select irrespective of the negligence rule. Therefore, if the problem is that courts do not have precise information on the parties' idiosyncratic characteristics, the solution is not necessarily comparative negligence Our point: cooperation-mimicking sharing In this paper, we claim that comparative negligence can serve as a buffer against erroneous due-care standards. Our argument is grounded in the idea that all-or-nothing 27 See also Edlin (1994) submitting that, under evidentiary uncertainty, due-care standards should be more lenient under contributory negligence than under comparative negligence. In our analysis, we assume that the due-care standards are given, so that they are not instrumental towards achieving efficiency. 15

32 rules (simple and contributory negligence) employ a stick (the payment of damages), which is applied entirely to one party. In contrast, comparative negligence shares this stick between negligent parties. Dari-Mattiacci and de Geest (2010) show that all-or-nothing sticks can be used to extract more effort from noncooperative parties than the amount of the stick itself, that is, sticks have a multiplication effect; the threat to punish is first applied entirely to one party and, if that party complies, it can be again entirely applied to the other party. Translated into an accident setting, this means that all-or-nothing rules can result in both parties taking care (for instance, $60+$60) in excess of the accident loss that care prevents (for instance $100). The multiplication effect can be countered by rules that share the stick between the parties. Sharing works as an implicit cooperation device. If parties could cooperate, they would agree to be negligent and share the loss between them (say, each party could pay $50). A rule that shares the loss between them (say, 50%) imperfectly mimics this cooperative agreement. In this paper, we draw on the multiplication principle in order to explain that, if the due-care standards are erroneously set, comparative negligence induces the parties to violate them and choose levels of precautions that advance social welfare. Our analysis differs from Cooter and Ulen's analysis in that we do not make any assumption about the size and effect of distortions in the care levels; moreover, we consider erroneous but certain standards, rather than uncertain standards as Cooter and Ulen do. 28 To be sure, we do not downplay the importance of evidentiary uncertainty; rather we think that it affects all rules in a similar way (Bar-Gill and Ben-Shahar 2003). Differently from Rubinfeld (1987), due-care standards in our model are erroneous not because there is heterogeneity in the population of injurers and victims, but rather because the court (or the regulator) makes mistakes in calculating the ex ante probability of accidents as a function of the parties' care levels. As a result, the due-care levels might be set too high or too low. There is a fundamental difference between information on the parties' idiosyncratic characteristics and information about the probability of accidents. The parties' idiosyncratic characteristics may be difficult to ascertain with certainty but concern factors that pertain to the accident that has happened; for instance, they relate to the costs 28 In a recent paper, Stremitzer and Tabbach (2009) also study the case of erroneous due-care standards in a setting different from ours; in their setting, precaution is unilateral, there is insolvency, and liability is proportional to the probability that the harm was caused by the injurer's negligence. Leshem and Miller (2009) investigate the performance of all-or-nothing rules versus damages proportionate to the degree of uncertainty in the defendant's liability. In their paper erroneous due-care standards play no role and precaution is unilateral. 16

33 of the parties' care or the magnitude of the harm. In contrast, information on the probability of accidents or, more precisely, to the mapping between the parties' many possible care choices and the probability (and, possibly, magnitude) of accidental harm concerns all the possible accidents that could have happened but, in fact, have not happened. Hence, if there is a problem with respect to the courts' inability to gather all relevant information, this problem is likely to be more serious when it comes to estimating the probability of accidents. 29 While lack of information on the parties' idiosyncratic characteristics may provide an argument in favour of any negligence rule, we show that lack of information about the probability of accidents casts an unambiguous vote for comparative negligence. Our proposal can be implemented by courts irrespective of a court's awareness that some information is lacking. The correction of errors made in setting due care is not a deliberate activity. Rather, relative fault can be easily implemented in all cases and it will make a difference only in those cases in which due-care standards are too high How to share A theory of comparative negligence should also provide guidance as to how to share the damage between the parties. Fairness and insurance theories offer easily implementable solutions. Prima facie, a 50% sharing is the preferred rule, 30 unless more information is available and a different sharing seems more appropriate. The filtering theory offers a simple recipe for a subset of cases (unilateral care with symmetric distribution of the costs of care): sharing should be equal to the probability that each party can avoid the accident. However, a more complex sharing is required in general. 31 All other theories offer little guidance as to how the court should pinpoint the optimal sharing of the loss between the parties, because the optimal sharing rule is a function of the specifics of the situation. 32 Showing that an optimal sharing of the loss is preferred to the all-or-nothing allocations does not imply that any sharing of the loss is better than the all-or-nothing 29 Haddock and Curran (1995, pp ) discuss a number of errors that the parties or the court could make, including errors concerning the probability of accidents, but conclude that to offset these errors courts would have to first understand the nature and the magnitude of the error and this is unlikely to happen. Therefore, no negligence rule is likely to be unambiguously superior. Our proposal does not require the court to assess the magnitude of errors. 30 For instance, Posner (2010, p. 223). 31 Dari-Mattiacci and de Geest (2005, p. 217). 32 See Orr 1991 and Cooter and Ulen (1986, p. 1092). Rubinfeld (1987, pp ) provides an example with a linear sharing rule based on a fixed sharing of the loss but does not expand on how a court could determine the exact sharing and what information this task requires. 17

34 allocations. Thus, if courts are not able or lack the information necessary to implement the optimal sharing, social costs may rise rather than decrease as a result of the introduction of comparative negligence. Therefore, a theory of comparative negligence based on the courts' lack of information concerning some crucial variables of the model should not rely on a sharing rule that, to implement, requires precisely the information that courts lack. Overly complex sharing rules or sharing rules that improve incentives only if finely tuned are unlikely to do the job. In contrast, the rule that we support, relative fault, can be implemented with information the court has and it requires less information than the determination of negligence. Case law and statutes contemplate various forms of sharing. Pure comparative negligence shares the loss according to each party's percentage of negligence; a modified form of comparative negligence bars recovery if the plaintiff's relative negligence exceeds a certain threshold (49% in some jurisdictions and 50% in other jurisdictions); finally, slight-gross comparative negligence bars recovery unless the plaintiff's negligence is slight and the defendant's negligence is gross. 33 However, the law is not very clear about how these percentages should be measured and what relative negligence means. Different interpretations have been proposed, that underscore the importance of such factors as the relative degrees of fault, the relative degrees of causal contribution to the accident, and the characteristics of the parties and their conduct. 34 The rule we propose, relative fault, shares the loss exclusively according to the parties' relative departures from due care An illustration of the results Consider a simple scenario in which care involves a discrete choice. An accident can be prevented if the injurer and the victim spend fixed amounts and, respectively, on care. If either or both of them fail to take care, an accident L will occur with probability. The accident should be prevented if ; otherwise the parties should not take care. An ideal liability rule should hold a party negligent for not taking care only in the former case. 33 McIntyre v. Balentine, 833 S.W.2d 52 (Tenn. 1992). See Grehan (1981), Barnes and Baeverstad (1982), Cooter and Ulen (1986, pp ) and Artigot I Golobardes and Gomez Pomar (2009, pp ). 34 Stanford v. Chevrolet Division of General Motors, 642 P.2d 624 (or. 1982); Watson v. State Farm Fire & Cas. Ins. Co., 469 So. 2d 967 (La. 1985); Restatement (third) of Torts: Apportionment of Liability 8 (2000). See also Prosser (1953, p. 481), Parisi and Fon (2004) and Edelman (2007) (distinguishing between absolute negligence and relative negligence; what we call relative fault corresponds to absolute negligence in Edelman's terminology). 18

35 While the parties have complete information, the court can only perfectly verify, and, but makes mistakes in assessing, which the parties anticipate. As a result, the court's probability estimate might be greater or less than the real. There are cases in which, meaning the court determines that care should have been taken (and failure to do so amounts to negligent behaviour) while in fact this is not the case. In other cases, the court might underestimate the probability of accidents and erroneously conclude that care should not have been taken. We will examine the ex ante incentive effect of court errors under simple, contributory and comparative negligence. (These results can be easily extended to strict liability rules). As illustrated in Figure 1, these rules only differ with respect to the apportionment of damages between negligent parties: (simple negligence), (contributory negligence) or (comparative negligence). 35 Figure 1. Apportionment of damages under negligence rules Injurer negligent Injurer non-negligent Victim non-negligent Injurer Victim Victim negligent Injurer s share Victim s share Victim If the court assesses, parties are non-negligent by definition because failure to take care does not amount to negligent behaviour in the eyes of the court. This generates incentives for both parties not to take care irrespective of the chosen negligence rule, whether or not care would be socially desirable. (The injurer does not take care because he bears no loss; in turn the victim does not take care because she cannot prevent the accident unilaterally.) If instead the court assesses, parties will be found negligent if they did not take care and the game has the form depicted in Figure 2. If this is a case in which the accident should be prevented,, the due-care standards set by the court are 35 Note that if the parties will be found non-negligent by the court regardless of whether they took care. Hence, an accident may occur even if both parties are non-negligent in the eyes of the court and the loss will be borne by the victim. In contrast, if the court will find the parties non-negligent only if they took care. Hence, if both parties are non-negligent there will be no accident and hence no damage to bear. 19

36 socially optimal and it follows from the well-known efficiency equivalence theorem 36 that the game has a unique Nash equilibrium where both parties take care under all negligence rules. Figure 2. The parties expected costs under negligence rules Injurer negligent Injurer non-negligent Victim non-negligent Victim negligent ( ) The interesting case arises when the court erroneously assesses the probability of accidents to be so high that care is deemed justified, while in fact care is not socially desirable ( ). In this case, the superiority of comparative negligence originates in the combination of two factors: efficient negligence and relative fault. If negligence emerges in equilibrium, then it is efficient negligence (a result that echoes Grady, 1998). Suppose that the outcome in which both parties are negligent is an equilibrium. This can be the case only if both parties prefer being negligent rather than unilaterally non-negligent ( and ( ) ), which implies that the sum of the parties' care expenditures is larger than the the sum of the parties' shares of expected damages, that is, of the total expected damages ( ( ) ). In turn, this shows that being negligent is socially desirable and thus, that negligence is efficient. 37 If negligence is efficient whenever it emerges in equilibrium, then a negligence rule that maximizes the chance of efficient negligence is to be preferred to other rules. A comparative negligence rule that apportions damages according to relative fault is such a rule. First assume that care is socially undesirable:. The goal is to induce both parties to be negligent. Multiplying both sides of the last inequality by and rearranging the terms, we obtain relative fault (. This shows that if damages are apportioned according to ) the injurer can be induced not to take care (and be negligent in the eyes of the court) whenever care is not socially efficient. (The same applies to the victim). Therefore, relative fault has the property of inducing efficient negligence. Substituting the inequality with a strict equality, one sees that relative fault is the only rule with this 36 See footnote 20 above and accompanying text. 37 In the next section we show that this result applies generally, hence also to unilateral negligence. 20

37 property. Simple and contributory negligence (along with comparative negligence rules based on other methods of apportioning damages) fail because they apply too much liability pressure on one party and too little on the other, as shown numerically in Example 2. In contrast, relative fault guarantees that the liability pressure is dosed optimally, so that both parties have enough incentives to be (efficiently) negligent Formal analysis Here, we consider the standard model of accidents. Accidents occur between two wealth-maximizing, risk-neutral parties: a victim and an injurer. Both the injurer and the victim can take care (, respectively) in order to reduce a monetary expected harm. The expected harm depends on both care levels and is expressed by ( ), which is assumed to be strictly convex in and, with ( ) and ( ) Setup and efficiency equivalence Two standards of due care, and are set by the court or a regulator. We examine three negligence rules: simple negligence, contributory negligence and comparative negligence. (The analysis of strict liability rules is analogous). If both parties abide by the due-care standards, the victim bears the harm. If one party abides by the due-care standard, while the other is negligent, the negligent party bears the harm. Finally, if neither party complies with due care the injurer pays a share of the damages and the victim bears the remaining portion,. We have simple negligence if : the injurer pays damages if he is negligent, irrespective of the behaviour of the victim. We have contributory negligence in the opposite case of : the injurer pays damages if negligent, but if the victim is also negligent the injurer's liability is waved. Finally, we have comparative negligence if, implying that if both parties are negligent, each of them bears a portion of the damages. Figure 3 shows the expected costs to be paid by the injurer and the victim respectively, under all three rules. Under all rules, damages awarded to the victim are assumed to be perfectly compensatory, that is, equal to the harm suffered. 38 Technical details about the equilibria of this game are provided in Appendix 2.1 Solution of the discrete game. 21

38 Figure 3. The parties expected costs under negligence rules ( ) As is common in the literature, we take the social cost to be the sum of the costs of care and the expected damage. The socially optimal levels of care are the levels of and that minimize this sum. Let and denote the care levels that minimize the social cost of accidents: [1] ( ) ( ) Assuming that and are positive, these socially optimal care levels are implicitly defined by the following first order conditions (subscripts indicate derivatives): [2] ( ) [3] ( ) If the regulator sets due care equal to the socially optimal levels (, and ), the game has a unique Nash equilibrium, where both the injurer and the victim take due care (, and ). Landes and Posner (1980, p. 539, fn. 51) and Haddock and Curran (1995, Sec. III) have demonstrated that this result holds under all three rules considered, that is, irrespective of the sharing. Rea (1987, Prop. 2) has shown that this result also holds true for ( ), that is, a sharing of the loss which is endogenously determined by the levels of care chosen by the parties. It follows that all rules are equivalently efficient. For ease of exposition, in the next section we first examine an exogenously given and later extend the analysis to an endogenously determined ( ). We now consider a situation in which the regulator erroneously sets due care too high or too low from a social welfare point of view. Due care can be erroneously set for different reasons. The regulator might inaccurately assess the relevant costs and benefits, the courts might be biased, the parties might erroneously interpret the legal standard, or, as we have emphasized in the previous sections, the court or the regulator may lack relevant information on the probability of accidents. We assume that the parties are informed about the standards of due care set by the regulator or can predict the standards of due case as will be set by the court. As we have done in Section 2.3, our strategy is to first show that negligence efficiently emerges in equilibrium and then to show that comparative 22

39 negligence with relative fault is the negligence rule that facilitates efficient negligence the most Equilibrium deviations from due care are efficient If the level of due care is different from the socially optimal level, the parties might take levels of care ( ), which are privately optimal for the parties but different from the due-care standards. The first question that arises is whether such deviations from due care reduce the social cost of accidents or not. We focus on purestrategy Nash equilibria. 39 The following Lemma shows that if the parties deviate in equilibrium from the due-care standards, the resulting social cost of accidents is less than it would have been had they abided by the due-care standards. Lemma 1. If due care is too high or too low, any equilibrium in which one or both parties deviate ( ) and/or ( ) yields less social cost than would result if both parties took due care: ( ) ( ). The intuition behind this result is that negligence rules are such that the party who decides to deviate internalizes all costs and benefits of his decision and hence his decision to deviate must be socially advantageous. This result implies that negligence efficiently emerges in equilibrium and thus should be somewhat encouraged. The following proposition shows that, since negligence efficiently emerges in equilibrium, if there are two equilibria, one of which is a compliance equilibrium and one a noncompliance equilibrium, then the latter is preferable because it yields lower social costs. Proposition 1. If there are two equilibria, one in which both parties are non-negligent, ( )and one in which both parties are negligent, ( ), then the latter yields less social costs of accidents: ( ) ( ). Corollary 1. If the outcome in which both parties are non-negligent is an equilibrium ( ), an outcome in which one party is negligent and the other is nonnegligent, ( )or ( ) cannot be an equilibrium. 39 Mixed-strategy equilibria are unstable. 23

40 Proposition 1 shows that whenever negligence emerges in equilibrium, it is also desirable from a social point of view. Corollary 1 generalizes the results of Proposition 1 by excluding the possibility that other equilibria arise Comparative negligence stimulates efficient negligence Since equilibrium deviations away from due care are always socially efficient, a liability rule that stimulates equilibrium deviations reduces the total social cost of accidents. The core of our argument is that comparative negligence is the rule that stimulates efficient deviations most frequently. We will examine several different cases and conclude that comparative negligence is either equivalent or superior to rules that burden one party only, such as simple negligence or contributory negligence. Proposition 2. If compliance is an equilibrium, comparative negligence yields lower social costs than simple and contributory negligence; if compliance is not an equilibrium, negligence rules cannot be generally compared Case 1: Compliance is an equilibrium If there is an equilibrium in which both parties take at least due care, then there cannot be a second equilibrium in which one party violates and the other complies (Corollary 1). This situation occurs for values of the due-care standards that are sufficiently close to the socially optimal levels. The only remaining possibility is a second equilibrium in which both parties are negligent ( ). Proposition 1 shows that if such an equilibrium exists then it is also desirable from a social point of view. A negligence equilibrium arises if both parties find it more convenient to bear a share of the damages than to abide by the due-care standard. Therefore, the injurer's share of damages should be high enough to prevent him from complying but not so high that the remaining victim's share falls below her compliance costs. The balance of the injurer's and victim's incentives to be negligent gives a range of possible values of the sharing that support the efficient negligence equilibrium: [4] ( ) ( ) An equilibrium in which both parties are negligent exists if and only if there is at least one value of that satisfies equation [4]. It follows that the condition in equation [4] is necessary and sufficient for the existence of a negligence equilibrium. An optimal sharing 24

41 rule is therefore one that implements a sharing such that it falls within this range of values. It is easy to see that rules that always implement (simple negligence) or (contributory negligence) are not very likely to fall within the range, and hence often forgo the possibility to induce a socially desirable equilibrium. In contrast, comparative negligence could be designed in such a way as to implement an appropriate value of within the range Case 2: Compliance is not an equilibrium In contrast with the previous case, here we consider the case in which compliance by both parties is not an equilibrium. This situation occurs for values of the due-care standards that are sufficiently greater than the socially optimal levels. We make a distinction between two subcases: 1) Negligence by both parties cannot be an equilibrium. Here we consider the subcase in which due care is such that the there is no that satisfies equation [4]. Then there cannot be an equilibrium in which both parties are negligent, irrespective of the liability rule in force. Since the emergence of equilibria in which one party is negligent and the other is non-negligent does not depend on the sharing between negligent parties, the choice between comparative negligence and other rules does not affect the social loss. 2) Negligence by both parties could be an equilibrium. Here we consider the subcase in which due-care standards are such that there exists a that satisfies equation [4], so that negligence by both parties could be an equilibrium. By hypothesis, compliance by both parties is not an equilibrium. There remain two other possible equilibria: one in which the injurer is negligent while the victim is non-negligent and another in which the injurer is non-negligent while the victim is negligent. Which one of these three possible equilibria is desirable from society's point of view depends on the levels of and and on the characteristics of the expected harm ( ), so that under different configurations a different rule could be desirable. Summing up, in case 1 (non-negligence is an equilibrium) comparative negligence can be designed so that it performs better than other rules, while in case 2 (non-negligence is not an equilibrium) the different rules cannot be generally compared. Proposition 2 describes a situation in which comparative negligence can induce more efficient care than alternative rules by allowing the parties to correct erroneously high standards of due care and take less but more efficient care. As Proposition 2 shows, this is 25

42 the case when two equilibria can coexist: an equilibrium in which both parties comply and an equilibrium in which both parties violate the standards of due care Apportioning damages according to relative fault So far, we have considered a sharing rule that is exogenously determined and announced ex ante to the parties. Even though it is possible to determine the optimal exogenous sharing rule, in practice this is unlikely to happen. The problem with an exogenous is that a regulator who makes mistakes in determining the due-care standards cannot be expected to set the sharing of the accident loss in an optimal way, so that parties can efficiently deviate from the due-care standards (although, in general, some intermediate levels of might still prove more efficient than simple or contributory negligence). Most commonly, the apportionment of the loss is done ex post in court, on the basis of an (ex ante determined) balance of the parties' respective negligence. This makes the sharing endogenous to the model ( ( )), with the natural assumptions that ( ) and ( ) (if a party raises his level of care, his share in the damages decreases). As we have already noted, the determination of the social optimum and the efficiency equivalence result in Section also hold true with an endogenous sharing rule. Likewise, the analysis in Section 2.4.2, which shows that deviations from due care are efficient when they arise in equilibrium, does not change with an endogenous sharing rule, thus all our results (Lemma 1, Proposition 1, and Corollary 1) remain valid. The analysis of Section can be easily adapted to an endogenous sharing rule by modifying the relevant range in equation [4] as follows: [5] ( ) ( ) ( ) Proposition 2 implies that if an equilibrium ( ) exists, in which both parties violate the due-care standards, and if the resulting apportionment of the loss falls in the range equation [5], then comparative negligence yields lower social costs than simple and comparative negligence. In turn, an equilibrium ( ) exists if there are values of the parties' care ) and ), such that ( ) ( ) ( ( )) ( ) 26

43 In this section, we examine the simplest and most intuitive endogenous comparative negligence rule, which shares the loss according to the parties' relative fault 40 : [6] ( ) ( ) ( ) Proposition 3. The sharing rule ( ) (relative fault) equal to the parties relative departures from due care is the only sharing rule that always falls within the optimal range given in equation [5] whenever negligence by both parties can be an equilibrium. The intuition behind this result is that relative fault apportions liability depending on the parties cost of taken care and on their outside options (taking due care). By doing so, relative fault balances the incentives of victim and injurer so that both parties prefer to be negligent when this is socially desirable due to excessive due-care standards A numerical example Here we present a numerical example illustrating the main results obtained above. In this example, we compare comparative negligence with simple and contributory negligence - an analogous example can be constructed to discuss strict liability rules. Consider the following simple function for the expected accident loss: ( ) with. Note that the parties' precautions are complements ( ). 41 The social welfare function is ( ) 40 See Artigot I Golobardes and Gomez Pomar (2009, pp ) for a classification of the possible sharing rules under comparative negligence. Note that this sharing rule has the desired properties: ( ), ( ), ( ) and ( ). This implies that the second order conditions for a minimum are satisfied for both parties: ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ( )) ( ) 41 This function has the desired properties: ( ) ( ) ( ) ( ) and ( ). The necessary condition for a minimum ( ( ) ( ) ( ( )) is ( ) ( ) which is satisfied because. The partial derivatives are not defined at or y (or both, for the cross partial derivative). For the first derivatives, the singularity can be removed by defining ( ) ( ) and ( ) ( ). Given that the second and cross-partial derivatives are well-defined in any neighbourhood of 0, this does not create problems for the validity of the internal solutions. 27

44 and it is easy to see that the socially optimal levels of care are (for any value of ). 42 Now assume that the parameter is and that both standards of due care are set at an excessively high level, say and. Both parties will be induced to comply (compliance is an equilibrium). Under the rules we consider in this example, the victim bears the full accident loss and his cost of care if the injurer is non-negligent, irrespective of his own level of care. Given that the injurer complies, the victim's cost is ( ), which is minimized by (all values are approximated); thus, the victim does not have an incentive to deviate. In turn, the injurer complies because his cost while complying is less that the cost he would bear if he did not comply:, for all The right-hand side of this inequality is minimized by and is at least equal to ; therefore, the injurer does not have an incentive to deviate. It can be easily verified that there is no other equilibrium in this game; 43 thus, the excessive due-care standards will inevitably result in excessive care by both parties (the victim takes even more care than the due level). The total social cost with compliance is ( ), which is larger than the social cost at the socially optimal levels of care, ( ). Let us now consider the more extreme case in which the standards of due care are set at an even higher level: and. Again, there is a compliance equilibrium in which both parties take at least due care. Similar to the previous case, the victim complies because his cost (if the injurer takes is minimized by. Likewise, the injurer complies because the due-care standard is less than the cost of unilateral noncompliance: ( ) ( ), with. The social loss due to the excessive due-care standards is even larger now, with a total cost of accidents reaching ( ). 42 The first-order conditions are, which yields the result. 43 The two equilibria in which a party is unilaterally negligent are ruled out by the Corollary. The case in which both parties are negligent can be ruled out by noting that a necessary condition for this equilibrium is ( ) ( ). This condition derives from summing up the inequalities in equations [7] and [8]. However, note that ( ). Thus, the outcome in which both parties are negligent cannot be an equilibrium. 28

45 Under simple negligence and contributory negligence, the compliance equilibrium is the only equilibrium of this game. 44 However, under a comparative negligence rule with, there is an equilibrium in which both parties violate the (excessive) standards of due care and take very low levels of care: 45 and This is an equilibrium because ( ) ( ) and ( ) ( ) ( ). Most importantly, this equilibrium, in which both parties violate the standards of due care, yields less social costs than the compliance equilibrium, ( ) although this is not the socially optimal level, it is close to it. The problem with a comparative negligence rule is that the sharing rule is crucial in order to obtain a welfare-enhancing noncompliance equilibrium. Therefore, let us now consider the case in which the sharing of the loss between negligent parties is endogenously determined as a function of the parties' relative negligence; that is, the sharing is ( ) as in equation[6]. There is a unique Nash equilibrium in which both parties violate the due-care standards. The levels of care chosen by the parties are and. These levels of care induce a sharing which is approximately ( ), 46 that is, close to the sharing considered above. Note, however, that with an endogenous sharing the parties' levels of care are very close to the socially optimal levels. This is due to the fact that an endogenous sharing rule improves the incentives of the 44 The two equilibria in which a party is unilaterally negligent are ruled out by the Corollary. The case in which both parties are negligent can be ruled out as follows. Under simple negligence, if both parties are negligent the injurer pays the full accident loss and his cost of care, thus this outcome could be an equilibrium only if ( ) ( ), where is equal to 0, because the victim faces no liability. Therefore the condition becomes, where also is equal to 0. However,, hence the outcome where both parties violate the standards of due care is not an equilibrium under simple negligence, because the injurer has an incentive to deviate. An analogous argument proves the this outcome is not an equilibrium under contributory negligence. 45 The first order conditions are ( ), which yields. Substituting, we have ( ) and ( ). 46 These values are obtained by means of a numerical simulation in the Mathematica 6.0 environment. The code is available with the authors. The exact values returned in the simulation are, and ( ) Given these equilibrium values, the totals costs for each of the parties when both are negligent are lower than if one party unilaterally deviates and takes due care: after rounding up, for the injurer, we have And for the victim we have ( ) [ ( ) ( ) ] ( ) [ ( ) ( ) ] 29

46 parties to take care and partially corrects for the dilution of the incentives due to the imperfect internalization of the accident loss. In particular, the levels of care obtained with an endogenous sharing rule are closer to the socially optimal levels than those obtained with an exogenous sharing rule; hence, the social loss is also less with endogenous sharing: ( ) ( ). Summing up, the endogenous sharing rule proposed above enables the parties to lower social costs by violating excessive due-care standards and instead choosing levels of care that are closer to the socially optimal levels Extensions Sequential care choices In the previous sections, we have focused on parties who act simultaneously. In some accident contexts, parties act sequentially: this means that a party might be able to choose his or her level of care after observing the other party's care choice. If the courts make no errors, both parties are induced to take the socially desirable level of care (Wittman, 1981). 47 If the standards of due care are too high, however, comparative negligence induces efficient negligence just as it did in the simultaneous case. 48 Consider again Example 2 and assume that Xavier moves first. Under simple negligence, if Xavier does not take care he is liable. Hence he will choose to take care and spend $30 instead of $100. Given that Xavier takes care, Yvonne will also take care, because her cost of care ($90) is less than the damages that she would bear ($100) otherwise. Consider now contributory negligence. If Xavier takes care, Yvonne will also take care. If Xavier does not take care, Yvonne's decision depends on the liability rule. Under contributory negligence, Yvonne faces a choice between taking care ($90) and bearing the damages ($100) and hence she will take care. Therefore, under contributory negligence, Xavier expects Yvonne to take care irrespective of what he does, and will find it more convenient to take care ($30) rather than paying damages ($100). Under comparative negligence, if both parties are negligent, Yvonne bears a share of damages equal to $75, which is less than her cost of care ($90). Thus, if she observes that Xavier is negligent, she will also choose not to take care. Anticipating Yvonne's reaction, Xavier will choose not to take care and bear his fraction of damages ($25) rather than pay 47 In the model developed by Shavell (1983) inefficiencies arise under some liability rules because the first movers choice of care affects not only the expected accident loss but also the probability that care by the second mover will be necessary. We do not consider this scenario here. 48 If due-care standards are too low, all rules perform in a similar way. 30

47 the cost of care ($30). As in the simultaneous case, in the sequential case both parties decide not to take care when this is efficient. The result is even more robust here because the game has a unique equilibrium: the equilibrium in which both parties took care disappears. (We could obtain the same result if we let Yvonne move first.) Note that, as in the simultaneous case, comparative negligence induces negligence only when this is efficient. If Yvonne's cost of care were equal to $60, Xavier's and Yvonne's liability shares would be and and neither party would find it convenient to act negligently. This outcome would be efficient because the sum of the parties' costs of care is now less than the damages Courts err with respect to the parties' costs of care We have argued that court errors are most likely to occur with respect to the probability and magnitude of the accident. However, in some cases the court might err with respect to the parties' costs of care. This might be the case because the court does not consider the cost of avoiding any inadvertent failure to take care by the parties (Grady, 1998). With wrong estimates of the parties' costs of care, the court will come to a wrong assessment of relative fault and hence might attribute an excessive or insufficient share of damages to either or both parties. Thus, comparative negligence might lose its error-correction feature. Nevertheless, it is plausible that in a large number of cases the courts will tend to under- or overestimate both parties' costs of care in a similar way, thereby preserving the balancing of relative fault unaltered. If this is the case, comparative negligence continues to have an error-correction function also when the court errs with respect to the parties' costs of care. Consider once more Example 2 and assume that the court correctly calculates damages ($100) but underestimates the costs of care by 33%; that is, the court believes that Xavier's and Yvonne's costs of care are $20 and $60, respectively. This error makes the court conclude that the accident should have been prevented, because $20+$60=$80 is less than $100, and thus that the parties should be found negligent if they did not take care. Under simple or contributory negligence, this error results in an inefficient outcome: the parties will take care. The error is innocuous under comparative negligence. The parties' shares of liability are for Xavier and for Yvonne. These are the same shares that the court would have obtained if the costs of care had been correctly estimated (see the text accompanying Example 2). Thus, under comparative negligence the parties will be efficiently negligent. The example demonstrates a general result: when the court 31

48 makes the same mistake in assessing both parties' costs of care, the balance of relative fault is unaffected and hence the error-correction function of comparative negligence remains valid Multiple tortfeasors So far, the narrative of this article has focused on accidents between an injurer and a victim, but one could easily extend the analysis to consider accidents with multiple tortfeasors, such as accidents with two injurers and one victim. If courts make no errors, all rules perform in the same way (Landes and Posner, 1980). Instead, if the courts make the kind of mistakes analysed in this article, the rule of contribution between the injurers should be modelled according to the relative fault principle in order to assure that excessive due-care standards do not lead to excessive precaution. Consider the following example. Example 4. An accident can be prevented if Xavier and Zachary (the injurers), and Yvonne (the victim) take care. Care taken by only one or two parties has no effect. Care costs Xavier $90, Zachary $60 and Yvonne $30. If an accident occurs, damages amount to $150. This is an accident that should not be prevented because $90+$60+$30=$180 is greater than $150. The comparative negligence rule combined with a contribution based on relative fault makes Xavier bear a portion equal to and Yvonne the remaining portion equal to, Zachary a portion equal to of the damages. Consequently, if all parties are negligent, they will bear shares of damages amounting to (Xavier), (Zachary) and (Yvonne). Each party bears a share of damages that is less than his or her cost of care and hence no party has an incentive to deviate from the negligence equilibrium. While apportioning damages according to relative fault reaches a desirable outcome, other rules may fail to do so when courts set due-care standards too high. Consider a rule of contributory negligence combined with a 50% contribution rule and assume that the 49 This result is true beyond the example. Assume that the parties' costs of care are and and that the court's estimates are and, where can be greater or less than 1. Xavier's share of liability is, which is unaffected by the court's error. The same applies to Yvonne's share. 32

49 court erroneously overestimates the expected harm and hence finds a party liable if that party did not take care. The victim bears the full accident loss if negligent and thus is induced to take precaution. The injurers share the damage equally if they are both negligent. Since $75 (Zachary's share of damages) is larger than $60 (Zachary's cost of care), Zachary will opt for taking care. Given that both Zachary and Yvonne have incentives to take care, Xavier will be left bearing full liability ($150), which exceeds his cost of care ($90) and induces him to take care Third-party externalities There may be situations in which excessively high due-care standards are socially desirable. This is the case when accidents generate additional negative externalities, such as harm to third parties other than the victim. Once harm to third parties is taken into account, the socially optimal due-care standards ought to be higher than those that would be obtained by only considering harm to the victim. However, harm to third parties does not enter the parties' strategic calculations: by hypothesis, the injurer does not internalize harm to third parties. This is precisely because such parties are not part of the lawsuit against him, so their harm will not be included in the calculation of the damages award. Likewise, the victim bears the costs of harm to herself - if she does not receive compensation from the injurer - but not the cost of harm to third parties. Since harm to third parties remains out of the calculations of the injurer and the victim, the socially optimal due-care standards are excessively high from the viewpoint of the injurer and the victim. The injurer and the victim will choose their levels of care irrespective of harm to third parties. Example 5. An accident can be prevented if both Xavier (the injurer) and Yvonne (the victim) take care. Care taken by only one party has no effect. Care costs Xavier $30 and Yvonne $90. If an accident occurs, damages amount to $100 (harm to Yvonne) plus $60 (harm to Zoe, a passive victim). Clearly, this accident should be prevented because the total loss ($160) is larger than the sum of the costs of care ($120). However, assuming that Zoe does not sue in tort, comparative negligence may make it easier for Xavier and Yvonne to violate the due-care standards and hence may dilute the incentives to take care. This is because the parties will only consider harm to Yvonne when deciding whether to take care. If faced with the 33

50 prospect of bearing damages equal to $75 and $25, respectively, when both parties are negligent, Xavier and Yvonne will not take care. In contrast, simple and contributory negligence would result in both parties taking care, which in this case is socially desirable. The important implication here is that in situations where harm to third parties is an important concern, and is taken into account in the due-care standards, comparative negligence is an undesirable solution. Other rules, such as simple or contributory negligence, should then be preferred. Why should Zoe not sue in tort? Negative third-party externalities may arise in all those cases in which not all victims are likely to sue the injurer. This may be because individual harm is trivially low - although aggregate harm might be very large, such as in some industrial accidents - or because not all victims have standing in court - such as in some environmental cases. In these cases, law and economics commentators have argued for the use of punitive damages (Polinsky and Shavell, 1988). Punitive damages do not remove the externality, because they only make the injurer, but not the victim internalize the harm to third parties. Yet, they remove the shortcomings of comparative negligence. Assume that damages are multiplied by 160%, for punitive reasons. As a result, under comparative negligence when both parties are negligent, the injurer pays 75% of $160 (the multiplied damages), that is, $120, which exceeds the injurer s cost of care and will induce him to take care. Given that Xavier takes care, Yvonne will also take care, since $30 is less than $100 (the harm to Yvonne). Thus, punitive damages correct the shortcoming of comparative negligence when there are negative externalities. Punitive damages also preserve the error-correction properties of comparative negligence, as the following modification of Example 5 shows. Example 6. An accident can be prevented if both Xavier (the injurer) and Yvonne (the victim) take care. Care taken by only one party has no effect. Care costs Xavier $135 and Yvonne $45. If an accident occurs, damages amount to $100 (harm to Yvonne) plus $60 (harm to Zoe, a passive victim). In this case, the accident should not be prevented because the sum of the costs of care ($180) exceeds the total harm ($160). Yet, with punitive damages, simple negligence induces the injurer to take care (because $135 is less than $160); consequently the victim will do the same (because $45 is less than $100). Hence simple negligence with punitive damages results in both parties taking care, which is not socially desirable. Instead, if 34

51 Yvonne is negligent, Xavier pays $120 (that is of $160), Yvonne earns $120 - $100 = $20 (what she receives from Xavier minus her actual harm) if she is negligent, and $160 - $100 - $45 = $15 (the full damages minus her harm minus her cost of care) is she is non-negligent. Since also her net surplus is larger when she is negligent, we can conclude that both parties will be negligent, which is the desirable outcome. This shows that comparative negligence has an error-correction property even with punitive damages. 50 Cooter and Porat (2006) identify a mirror-image case, in which the externality to third parties is positive. For instance, medical doctors' liability only considers the downside of their activity: the possibility that a medical treatment results in harm to the patient. However, there might be a relevant upside that is never internalized, as when the doctor takes some risks in order to save the patient's life. The doctor may bear liability if the patient dies but is not rewarded is the patient survives. This asymmetric treatment of otherwise symmetrical events generates a positive externality. In our framework, this results in the desirability of lowering due-care standards below the levels that would be optimal for the parties to the dispute only. This scenario of positive externalities leaves our analysis unchanged. If the standards are accurately set, all negligence rules perform in the same manner, while comparative negligence may be preferable if errors are taken into account Litigation rate An additional factor to consider is the effect that the choice of negligence rules has on administrative costs. As explained above, comparative negligence leads to more negligence than other negligence rules in the case of erroneous due-care standards. More negligence may mean more harmful accidents or more frequent accidents. Our analysis is based on the consideration that, if the due-care standards are too high, by hypothesis more harmful or 50 This result is true beyond the example. Assume that the parties costs of care are and, harm to victim is and harm to a third party is. Assume further that damages plus punitive damages are equal to. If Yvonne is negligent, Xavier will prefer to be negligent if ( ), that is, if his liability share is less than his cost of care. The latter inequality can be rewritten as, which guarantees that the injurer will choose negligence if and only is this outcome is socially desirable (the sum of the costs of care exceeds the harm). Likewise, if Xavier is negligent, Yvonne will prefer to be negligent if ( ) ( ), that is, if the harm she suffers, minus the damages paid by Xavier when both parties are negligent is less than her cost of care plus the harm she suffers, minus the damages paid by Xavier when he is unilaterally negligent. The latter inequality can be rewritten again as, which guarantees that also the victim will choose to be negligent is and only is this outcome is socially desirable. 35

52 more frequent accidents can actually result in an increase in social welfare because the additional costs to society are more than compensated by reduced costs of care. Nevertheless, if comparative negligence results in more harmful or more frequent accidents, one could associate this outcome with higher litigation rates and hence higher administrative costs. The relationship between negligence rules, the harmfulness or frequency of accidents, and litigation is complex. On the one hand, more harmful accidents raise the stakes of the lawsuit and parties might be more prone to litigate than to settle, a result that follows from the standard model of litigation. Comparative negligence, by inducing parties to take less care, may induce more harmful accidents and hence higher litigation rates. In turn, the larger costs for the system might more than offset the efficiency gains in terms of comparative negligence. The case of more frequent accidents is even more straightforward. If more accidents occur, there will be more disputes and hence more litigation in absolute terms, even if the litigation rate does not increase. On the other hand, de Mot (2013) shows that comparative negligence may reduce the stakes of the dispute compared to rules of simple and contributory negligence, in which the outcome is binary: a party either pays full damages or does not pay anything. In contrast, under comparative negligence parties litigate over marginal changes of the damages to be awarded, and hence may be prone to settle more often or to invest less in litigation. This results in lower administrative costs per accident. On balance, the effect of comparative negligence on the total administrative costs is ambiguous Conclusion In this paper, we have advanced a novel formal theory to explain why comparative negligence is such a frequently used rule. We have shown that comparative negligence can serve as a buffer against erroneous due-care standards, by inducing efficient violations more frequently than other negligence rules. What distinguishes our approach from previous literature is that we do not compare the social welfare resulting from the different rules in general, but rather identify those cases in which comparative negligence could make a difference by inducing parties to be (efficiently) negligent, who would otherwise follow due-care standards. We then show that in the remaining set of cases, negligence rules cannot be ranked and the effects of social welfare depend on the circumstances. 36

53 Our analysis formalizes the notion of efficient negligence developed by Grady (1998) 51 in the context of unintended violations of the negligence standard. In Grady s analysis, preventing all unintended violations of the negligence standard is too costly and hence certain violations occur efficiently. Grady argued that comparative negligence, by splitting the loss between the parties stimulates efficient violations of the negligence standards. Although we have not formally modelled unintended violations, our framework captures the idea that while parties might be aware of the inability to comply all the time, courts might be unable to assess the costs and benefits of inadvertent violations and hence set the due-care standards at a too high or too low level. 52 As we have shown, if parties have an informational advantage over courts, comparative negligence gives parties incentive to be efficiently negligent. The core of our argument is that, if due-care standards are too high, then it can be socially desirable that the parties violate such standards. Comparative negligence, with its low-powered incentives, is the rule that most often induces such efficient violations. Empirical studies have shown that comparative negligence does, in fact, induce lower levels of care than contributory negligence (Sloan et al., 1995; White, 1989). In addition, we have shown that if due-care standards are too low, all rules induce the same levels of care. Our analysis also offers support for a specific determination of the apportionment rule (relative fault), which courts should use to share damages among non-negligent parties both when there is a single injurer and with multiple tortfeasors. Finally, our theory offers a testable hypothesis concerning the emergence of comparative negligence: the emergence of comparative negligence in relatively recent times could be due to the difficulty for courts to determine the due-care standards in an increasingly complex world. 51 See also Grady (1984, 1990). 52 As we have explained in Section 2.2, our model does not relate to the fact that a party might notice that the other party have failed to take care. 37

54 38

55 3. A Comparative Review of the Protection of Traditional Knowledge a Theory of Three Approaches ABSTRACT This paper presents a new dataset of existing legislation on the protection of traditional knowledge, as it is reported to the WIPO. A theory of three approaches is introduced, explaining that the legislation that protects traditional knowledge can be subdivided in three distinct approaches, each with specific characteristics. The economic empowerment approach, through which the legislator seeks to create opportunities for the economically weaker indigenous groups in society to monetize on their traditional knowledge. The preservative protection approach, in which the legislator seeks to preserve folklore for the future and for the benefit of the country as a whole. And the cultural integrity approach, used to prevent the offensive and inappropriate use of sacred culture, historical objects and authentic new traditional knowledge products. A statistical analysis shows that the theory of the three approaches explains up to 81% of the variation found in the legislation on traditional knowledge reported to WIPO. 39

56 3.1. Introduction At the July meeting of the WIPO 53 intergovernmental Committee on Intellectual Property and Genetic Resources, Traditional Knowledge and folklore (IGC) the members of said committee have agreed to recommend the renewal of their mandate. The IGC s mandate calls, amongst others, for the creation of instruments for effective protection of Traditional Knowledge (TK). It seems that for such a task it would be instrumental to have an overview of the legislative instruments that currently exist for that purpose. TK is a broad concept that encompasses objects, stories, rituals and traditions. It holds cultural, economic and historic value, which explains why legislators are looking for ways to protect and realize these valuable assets. Such protection can be difficult when the forms of TK do not fit into existing concepts such as intellectual property, tangible property etcetera. This search for forms of legislation is still ongoing, as various legislators have found different forms of legislation to protect TK. This paper provides such an overview of existing legislation regarding the protection of TK, with the exclusion of Genetic Resources. All legislation reported to the WIPO is listed in a database and categorized with respect to its legal content. The categorization includes, inter alia, which forms of TK are protected, whom they are protected for and what they are protected from. A theory of three approaches is introduced, explaining that the existing legislation protecting TK can be split up in three approaches, each with distinctive characteristics. The economic empowerment approach, which seeks to create opportunities for the economically weaker indigenous groups in society to monetize on their TK. The legislation creates the possibility for indigenous communities to pursue active commercial interests for their traditional medicinal knowledge. South American countries with a relatively high GDP, and Asian countries with a relatively low GDP are likely to follow this approach. The preservative protection approach seeks to protect folklore for the benefit of the state or the country as a whole. This approach is found a lot in African countries, and outside of Africa one is more likely to find it in the legislation of countries with a relatively low GDP. The cultural integrity approach seeks to prevent the offensive and inappropriate use of sacred culture, historical objects and authentically created new objects. Legislation following this approach is most likely to be found in Oceanian countries with a high GDP, North American countries with a high GDP, or countries with a high GDP in general. 53 World Intellectual Property Organisation. 40

57 Statistical analysis shows that the theory of the three approaches explains up to 81% of all variation in the legislation that was reported to protect TK around the world. Legislators can follow either one of these approaches or can combine more than one approach in their legislation regarding the respective forms of TK. On what TK exactly entails one could have an endless discussion. In fact, a discussion on a formal definition of TK is currently on-going within the WIPO framework. Various suggestions have been made for a definition, and the current debate, amongst others, focuses on taking into account the purpose for the definition (intellectual property forms of protection, or also other policy contexts such as the preservation of cultural heritage, promoting biodiversity, or a human rights context), but it also focuses on the aim for a general coherence and compatibility between various policy instruments. For an elaborate portray of the on-going discussion, see WIPO Traditional Knowledge - operational terms and definitions 54. For the purpose of this paper I will use one of the definitions as issued by WIPO 55 : 'Traditional Knowledge' refers to tradition-based literary, artistic or scientific works; performances; inventions; scientific discoveries; designs; marks; names and symbols; undisclosed information; and all other tradition-based innovations and creations resulting from intellectual activity in the industrial, scientific, literary or artistic fields. Tradition based' refers to knowledge systems, creations, innovations and cultural expressions which: o have generally been transmitted from generation to generation; o are generally regarded as pertaining to a particular people or its territory; and o are constantly evolving in response to a changing environment. In the list of examples of categories that TK could include one example is, expressions of folklore in the form of music, dance, song, handicrafts, designs, stories and artwork. So expressions of folklore (folklore hereafter), are considered to be a part of TK, under the WIPO definition. This definition expressly excludes from the description: 54 WIPO Intergovernmental Committee on Intellectual Property and Genetic Resources, Traditional Knowledge and folklore, Traditional Knowledge Operational Terms and Definitions, 20 May 2002, WIPO/GRTKF/IC/3/9. 55 WIPO, Intellectual Property Needs and Expectations of Traditional Knowledge Holders: WIPO Report on Fact-Finding Missions on Intellectual Property and Traditional Knowledge ( ), April as referred to in WIPO Intergovernmental Committee on Intellectual Property and Genetic Resources, Traditional Knowledge and folklore, Traditional Knowledge - Operational Terms and Definitions, 20 May 2002, WIPO/GRTKF/IC/3/9. 41

58 ( ) items not resulting from intellectual activity in the industrial, scientific, literary or artistic fields, such as human remains, languages in general, and other similar elements of heritage in the broad sense. Whenever the term indigenous peoples, or indigenous communities is used in this paper, I will do so with the definition as made by the ILO 56 in mind. Indigenous peoples are defined as: ( ) peoples in independent countries who are regarded as indigenous on account of their descent from the populations which inhabited the country, or a geographical region to which the country belong, at the time of conquest or colonization or the establishment of the present state boundaries and who, irrespective of their legal status, retain some or all of their own social, economic, cultural and political institutions. From the given definition of TK it already becomes clear that the concept is quite broad and vague. This is done in the WIPO context, so as to not ex ante exclude any particular expression from the definition, before a clear framework of Intellectual Property (IP) protection is created. This broad scope of TK has the effect that the forms of expressions that are considered to fall under TK can vary a lot. To demonstrate the wide range of knowledge that falls under this broad concept of TK, examples 57 include not only dances, music and ceremonies, but also games, folk tale and mythology. It also includes tangible expressions such as art, instruments and architectural forms, but also intangible concepts such as spirituality, ethics and moral values. Traditional measures for food preparation, agriculture and conservation, and also practices of health care and education are mentioned as well. The difficulty with the protection of TK, lies in the fact that it does not fit into the common concepts that we use to protect forms of knowledge. Gervais (2005) provides a general analysis on how well the concepts of IP fit with the concepts of TK in a 56 International Labor Organization Convention 169, Convention Concerning Indigenous and Tribal Peoples in Independent Countries, June 7, 1989, art 1(b). 57 The given examples are taken from the following sources: UNESCO WIPO Model Provisions for National Laws on the Protection of Expressions of folklore Against Illicit Exploitation and Other Prejudicial Actions of 1982, section 2; UNESCO Recommendation on the Safeguarding of Traditional Culture and folklore of November 15, 1989, adopted by the General Conference at its 25 th session in Paris; UNESCO Symposium on the Protection of Traditional Knowledge and Expressions of Indigenous Cultures in the Pacific Islands, Noumea, February 1999; and Indigenous Knowledge, Local Pathways to Global Development, Marking Five Years of the World Bank Indigenous Knowledge for Development Program, Knowledge and Learning Group, Africa Region The World Bank,

59 compatibility analysis. Current IP concepts, like copyrights, trademarks and patents generally apply to concepts with the following three characteristics (Gervais 2005, p. 141): An identifiable originator, like and author or inventor; An identifiable work, like an invention or a sign; Defined restricted acts. As the examples of TK - mentioned above - already suggest, TK does not always show a comfortable fit with these three characteristics. TK is not static, it is ever evolving, and therefore it is sometimes hard to specifically describe and identifiable work. Each tribe of a certain indigenous people can have their own variation on a traditional game or ceremony, even though they are by the indigenous peoples themselves considered to be part of the same TK. This also leads to the problem that the originators of TK are seldom identifiable, as TK is often passed on and evolved over generations, as one could imagine is the case with folk tale and mythology, but also with practices of healthcare and food preparation. The literature on the protection of TK can be roughly categorized in three groups. The first discusses the political process and the challenges that are being faced by international undertakings to protect TK. The second group of literature zooms in on specific forms of TK, and how they can be protected. The third group of literature consists of case studies, comparing various approaches that countries have taken to protect TK. As stated before, in the recent past an increasing interest in the protection of TK can be seen in international arenas (Yu 2003). In the WTO 58, TK has emerged as an important issue on the agenda (Bodeker 2003). The CBD 59 and the IUPGR 60 are examples of international agreements that such increased interest can lead to. However, it is believed that a divide exists between developed and developing countries in the way this increased interest shows itself (Ntambirweki 2001, and Ghosh 2003). Developing countries have become more proactive in advancing TK as an issue to be dealt with on international forums 61. Developed countries seem to be more inclined to either not consider TK such a pressing issue, or to deal with TK-protection in their own domestic legal systems 62. Examples of literature regarding TK protection focusing on the challenges an international approach to the protection of TK would face (for example Arewa 2006, Yu 58 World Trade Organisation. 59 Convention on Biological Diversity, Rio de Janeiro, 5 June 5, The International Treaty on Plan Genetic Resources for Food and Agriculture, of the Food and Agriculture Organization of the United Nations., Also known as the International Seed Treaty, adopted by the FAO Conference on 3 November For example: Venezuela in WT/GC/W/282 at the WTO. 62 For example: Australia and the US: see Grad, R. (2003) p

60 2003 and 2008). McManis (2003) Provides a description of ongoing international initiatives, exemplary for the so called global thinking regarding TK protection. Dutfield (2001) goes as far as to state that TRIPs is not an appropriate forum to develop TK protection. Ghosh (2003) discusses three case studies, and concludes that in a mix of domestic and international law IP can be used strategically to protect what he calls traditionally subordinate groups, through the definitions of basic IP concepts such as prior art and novelty. He also points out the threats of appropriation that IP law can pose to TK. Gupta (1999) suggests regional cooperation between India, Brazil and South-Africa to find a protection for biological TK and traditional medicine, while mixing the informal traditional constitutional contexts in with the formal western institutions to develop policy in this regard. Most literature on the protection of TK does not enclose all forms of TK. The bulk of the literature focusses on Knowledge regarding Biological TK, and/or on traditional medicine. Possibilities for sui generis rights for genetic resources and Biological TK are explored by Cottier and Panizzon (2004), Downes (2000) and Halewood (1999). The same is done by Bodeker (2003) but for traditional medicine. Another direction of the literature discusses case studies of specific countries or projects. Grad (2003) compares the experiences of the US and Australia regarding the different paths the development towards expanding of their IPR systems have taken, to rectify inequality for indigenous peoples. In Australia this development initiated in court rulings, spilling over into legislation. In the US the legislator initiated this process in a response to perceived injustices in courts. Kutty (1999) compares the national experiences of India, Indonesia and The Philippines with respect to the protection of folklore, although he uses a broader definition of folklore than is done is this paper, including for instance arts and crafts. Kuruk (1998) describes various national legislations on the African continent regarding the protection of folklore, and discusses how unprotected this African folklore is in the US. Frankel (2007) explores the opportunities in trademark law to provide protection for TK, and compares the New Zealand experience with the US experience. McManis (2003) analysis a case study of the ICBG 63 -Peru Project as exemplary for a more novel way of using IP to protect TK. The cooperation of 5 organisations, 3 universities, a corporate party and an organisation of indigenous peoples, leading to a combined patent application regarding traditional plant knowledge of the Aguaruna people of Peru. 63 International Cooperative Biodiversity Group. 44

61 The literature regarding the national experiences is very educational, as they zoom in on a small group of countries. They give an elaborate analysis of the legislation, the TK and the challenges faced by the parties involved in the provision of protection of TK. These analysis give an in depth, but still only fragmented view of the existing legislation on the protection of TK. What is currently still missing is an overview of these national experiences. A birds-eye-view if you will, of the form and focus of national legislation. That is where this paper comes in. Firstly by giving an overview of legislation on the protection of TK, as reported to the WIPO, and analysing the directions that the various national legislators have chosen. Secondly this paper looks for similarities and differences in the legislation regarding the forms of TK that are protected, the people whom the TK is protected for, and what threats the TK is protected from. The analysis shows that the existing legislation protecting TK can be split up in three separate approaches, each with their distinctive characteristics. In section 3.2 a description of the database consisting of the categorized Legislation on the protection of TK is given, together with an explanation of how the database was constructed. Section 3.3 provides a descriptive and statistical analysis of the database, explaining and describing the three distinctive approaches of the national legislators. Section 3.4 concludes Legislation on the protection of TK description of the database As mentioned in the introduction, the various forms of TK do not always fit into IP concepts, such as trademarks, copyrights or patents. This can lead to unwanted situations. An example, taken from Gervais (2005), deals with commercial sound recordings using traditional music from indigenous people. When the original composer of such traditional music within the indigenous people cannot be identified, the music is not protected by copyright. A situation can then occur where a part of such traditional music is used in (a part of) new commercial sound recordings, by an artist who does not belong to this indigenous people. This new commercial piece of music can then be protected by copyright, as long as it fits all criteria. This leads to the situation that someone belonging to the indigenous people, may be prevented from using their own culture s music in a commercial way, if it is considered to infringe on this newly established copyright. It seems obvious that this cannot have been the goal of the IP laws in place, or at least that this is a worrisome situation. There are various ways in which legislators have decided to deal with this discrepancy. In this section a dataset is introduced that gives an overview of 45

62 the forms in which states have provided protection of Traditional Knowledge in their domestic legislation. In most countries it has been recognized that there are various forms of TK that do qualify for normal IP protection, as long as they fit the three characteristics that are mentioned in the section 3.1. An indigenous artist s painting can be protected by copyright. An indigenous people can trademark a sign or emblem, as long as it fulfils the three characteristics. Most countries that provide IP protection provide such a form of protection for TK. These forms of IP protection are often based on various guidelines and international treaties, such as the Berne Convention 64, the WCT 65, The TRIPs 66 agreement, and the WPPT 67 to name a few. In this section, however, I will focus on the protection that goes beyond these frameworks. The TRIPs agreement creates a framework of minimum standards for IP protection that every member has to uphold. The agreement is a result of the WTO negotiations in the Uruguay round. These international standards, however, require domestic legislation for their implementation. A certain degree of freedom in the way these standards are implemented is left to the discretion of the national legislators. This results in a situation where the WTO members each have their own domestic legislation, all meant to implement the same standards as set in TRIPs. An added diversification to this format is that the standards are minimum standards, and that member states are allowed to set their national standards of protection higher, if they wish to do so. In theory this results in all member states providing at least the minimum standards of IP protection, and some member states providing higher levels of protection. It does make sense, therefore, when addressing the status of legal protection of TK, to take the position of TK in the TRIPs treaty into account. With the creation of the dataset an overview is given of the existing domestic legislation of TK protection around the world, and a comparative analysis is done of the contents of these legislations. In section the dataset, and the way the data were gathered are explained. Section explains the qualitative categorization of the data. 64 Berne convention for the protection of literary and artistic works, as revised in Paris, July 24, World Intellectual Property Organization Copyright Treaty, Geneva, December 20, Agreement on Trade Related Aspects of Intellectual Property Rights, Annex 1C of the Marrakesh Agreement Establishing the World Trade Organization, Marrakesh, April 15, World Intellectual Property Organization Performances and Phonograms Treaty, Geneva, December 20,

63 Selection of the data This section gives a brief explanation of how the database was composed. Which legislative texts were included, and how the legislation was obtained. Listed states: All member-states of the WIPO are requested to report their legislation on the protection of TK. The states that have been listed in this overview are all states that have indeed reported to have such domestic legislation on TK to WIPO. Some states that are not members of WIPO have reported legislation on TK to WIPO, and are therefore also enclosed in this overview. These states are all island nations in the Pacific Ocean: The Federated states of Micronesia, Palau and Vanuatu. 68 Membership of both the WIPO and the WTO, as of April 2012 are listed for the states in the database. Listed Legislation: The legislation on which this listing is based is the legislation that states have reported to WIPO. And whenever in a text there is a reference to a specific other legal document that document has as well been listed in this overview. In the legislation that was provided by Ecuador, a reference is made to a law containing a sui generis right for collective IPR s of indigenous communities, that is to be created 69. At the moment of writing no such right had been implemented yet 70. Hence no such sui generis right is recorded in the database. Legislation that contained no reference to TK in any form as defined above has been left out of this overview. For the framework of this overview the definition used for TK is the same as mentioned in section 3.1. The WIPO definition of TK also encloses genetic resources, such as agricultural, biodiversity-related, and ecological knowledge. In WIPO genetic resources are both discussed as a part of TK 71, but the way it is referred to as separate from TK (often in the phrase Genetic Resources, Traditional Knowledge and Folklore ) shows some ambiguity on whether or not genetic resources should be considered a part of TK. Genetic resources are a form of TK that is regulated differently from other forms of TK. Living organism have qualities that TK does not necessarily have. 68 Neither of these states were members of the WIPO at the moment of the creation of the database. However Vanuatu has become a member of the WIPO as of March Law on Intellectual Property of Ecuador, 1998, art As of 11 th of August WIPO, Intellectual Property Needs and Expectations of Traditional Knowledge Holders: WIPO Report on Fact-Finding Missions on Intellectual Property and Traditional Knowledge ( ), April as referred to in WIPO Intergovernmental Committee on Intellectual Property and Genetic Resources, Traditional Knowledge and folklore, Traditional Knowledge - Operational Terms and Definitions, 20 May 2002, WIPO/GRTKF/IC/3/9. 47

64 Mainly migratory qualities of certain species originally prompted the protection of particular species through treaties, as early as the first half of the last century 72. The vast body of international treaties that has come into existing since then, arguably with The Convention on Biological Diversity (CBD) as a relatively recent climax in 1992, and its advanced evolvement compared to other international undertakings to TK protection shows that even if genetic resources would be considered a part of TK, they could be seen as an odd one out among the various forms of TK that exist. Therefore, to prevent confusion on the matter, all knowledge related to genetic resources, biodiversity, ecological and agricultural knowledge are seen as separate concepts form the other forms of TK and will be referred to with the broader term biological knowledge. There are many similarities to be found between the existing approaches to biological knowledge protection and the existing and possible approaches to TK protection, and much can be learned from the former with respect to the latter. However the protection of the two types of knowledge have shown very different patterns in their historical developments. The development of biological knowledge protection has historically developed top-down, from international treaties to domestic regulation, and the protection of TK has shown a bottom up development, and arguably has come up as an issue more recently. This has led to the exclusion of biological knowledge from the current analysis. Whenever a domestic legislation enclosed a protection of biological knowledge, it is marked in the dataset, but the protection is not categorized according to the qualifications that the other TK protection is categorized. States classification: The reporting by the states to WIPO also includes a classification of the legislation, as it is used by WIPO. These qualification are called Traditional Knowledge, Traditional Cultural Expressions, and Genetic Resources. WIPO does not provide any conclusive definitions on these terms, which leads to differing interpretations and confusion in the use of this terminology. Therefore it is not used in this paper. For a further discussion of the ambiguity of the use of these WIPO terms see appendix 3.1. Due to the definition that is being used in the listing of this database, both legislation that was reported to regard Traditional Knowledge and Traditional Cultural Expressions has been taken into account in this overview. 72 Early examples, taken from Bodansky, International Law and Protection of Biological Diversity, 28 Vanderbilt journal of Transnational Law, 623, 1995, footnote 3 include: The Convention for the Protection of Migratory Birds, August 16 December 8, 1916, between the USA and the UK, 39 Stat. 1702, The Convention for the Regulation of Whaling, September 24, 1931, 49 Stat. 3079, 155 L.N.T.S

65 The legislation that was reported to regard Traditional Knowledge and/or Traditional Cultural Expressions, but in fact only provided protection for biological knowledge has been recorded as such. Benchmark TK protection: Although not all WIPO Memberstates, and hence not all states listed, are Member to the TRIPs agreement, the minimum standards of TRIPs have been used as a benchmark to report the legislation. 73 Only legislation that provides any form of protection of a higher standard than the minimum standards of TRIPs have been listed. For the minimum standards only the forms of protection have been taken into account. The term of protection have not been taken into account. For example six countries 74 had registered to have domestic legislation protecting Traditional Cultural Expressions. Each of these six countries registered their general IP laws on copyrights, patents, trademarks etcetera. Neither of these six IP laws provide protection beyond the TRIPs standard. This means that the TK can only be protected as long as it fulfills the three requirements mentioned in section 3.1, such as an original painting that was made by an identifiable artist, not that long ago that the term of protection has ended. In the legislation of Macedonia this is explicitly mentioned 75, as in the other countries legislation it can only implicitly be deduced from the text. The legislation of these countries was marked as not exceeding the minimum standards of TRIPs Categorization The content of the legislation has been categorized. In this section the classification system is discussed that was used for the categorization. As the definition of TK shows, it is a broad term, and encompasses a lot of different forms of knowledge. The states that have provided for some sort of TK protection have done so, but for varying forms of TK. To distinguish between the various forms of TK a categorization has been made. In the database the following categories of TK have been identified: 73 At the moment of construction of the database, April 2012, There were 38 states that were a member of the WIPO, but were not a member of the WTO, and hence of the TRIPs agreement. 74 Fiji, Macedonia, The Federated states of Micronesia, Mongolia, Ukraine and the United Kingdom. 75 Art 6, Law on Copyright and Related Rights, of September 12, 1996, as amended by the Law of January 22, 1998, Macedonia. 49

66 Traditional medicine: The knowledge with respect to medication as well as techniques used in medicine. Authentic new objects (newly and authentically produced objects): Handicrafts, clothing, art and other tangible objects. Historical TK objects: TK objects that are considered historical artifacts due to the time when they were produced and used. Sacred culture: Religious or sacred places, expressions or rituals. Folklore: Traditional cultural expressions such as stories, songs, dances. Designs: Patterns used in handicrafts, art or clothing. Non-TK protected: Knowledge that does not fall under the TK definition used here, but has been protected by some of the reported regulations entails: o Biological knowledge: such agricultural knowledge, ecological knowledge and biodiversity related knowledge. Biological knowledge is excluded from this listing of TK. However certain domestic regulations protect not only the knowledge that falls under the definition of TK that is used here, but also provide some form of protection for biological knowledge at the same time. Although for the purpose of this categorization this is not considered TK protection, it is recorded when a state has provided such protection. o Other: Other forms of TK that have been mentioned are: Human remains 76 Animals 77 Geographical indications 78 Living individuals holding knowledge that is considered TK 79 Although these concepts do not fall under the definition of TK as used in this analysis, they are considered to be TK by the states reporting their legislation as laws protecting TK. Which is why they are listed under the other category. Rightholders: Answers the question who own the intellectual property of the TK, and who reaps the (economical) benefits of that TK. These are categorized as either 76 In the Aboriginal and Torres Strait Islander Protection Act, 1984, Australia, and the Native American Graves Protection and Repatriation Act (NAGRPA), 1990, USA. 77 Protection of Cultural Heritage Act, as amended by Act no. 8346, Apr. 11, 2007, Republic of Korea. 78 Decreto Legislativo N 1075: Decreto Legislativo que aprueba Disposiciones Complementarias a la Decisión 486 de la Comisión de la Comunidad Andina que establece el Régimen Común sobre Propiedad Industrial, Peru. 79 In the Historical and Cultural Preservation Act, 1995, Palau. 50

67 the state (or a state organ or agency), indigenous communities as a group, or individuals. Who instigates: Answers the question who has to take the initiative to get the TK protected. These also are categorized as either the state (or a state organ or agency), indigenous communities as a group, or individuals. The amount of action that is required from the holders of TK can determine the extent to which the possibilities of TK protection are actually brought in practice. Protected interests: The types of protection provided is split up in three categories: 80 o Active commercial interest: allowing parties to benefit commercially from the economic advantages that can stem from TK; o Defensive commercial interests: preventing parties other than the party appointed as the rightful owner to benefit commercially from the TK; o Ethical concerns: to avoid offensive use of TK, or to give recognition to the source of the TK used. GDP: in the analysis the GDP of the states is taken into account. For these data the IMF 2010 GDP per capita PPP in international dollars is used. For two states these data were not available. For Palau and The Federal states of Micronesia the Worldbank 2010 GDP per capita PPP in international dollars is used. Region: in the database the continents the countries belong to are listed to look for regional preferences. The continents are grouped according to the seven continent model of the United Nations Geoscheme. Countries that have territory spanning overland continental boundaries are listed to belong to both continents 81. Countries with overseas territories in other continents then their mainland are only listed as belonging to the continent of their mainland Analysis The database provides us a more ordered and organized overview of the contents of the national legislations protecting TK. Taking a closer look at the data provides insights in the similarities and differences between the different countries with respect to their legislations. In section a comparative analysis is given based on the birds-eye-view 80 This classification is taken from WIPO/GRTKF/IC/3/10, paragraph These countries were Egypt (Africa and Asia), and Azerbaijan, Georgia, Kazakhstan, Russia, Turkey (Europe and Asia). 51

68 the database provides. This so called eyeball analysis discusses the descriptive statistics of the database. Section discusses regional preferences regarding the protection of TK, and in section a theory of three approaches is introduced regarding the underlying preferences of legislators for the protection of TK, and how that links to the legislative choices they made. In section a statistical analysis is used to verify and fine-tune this theory Reported legislation The first thing that stood out during the building of the dataset, were the discrepancies between the reported content of the legislation and the actual content. The reporting states labelled the reported legislation as either regarding genetic resources, TK or traditional cultural expressions (folklore). As can be seen from the WIPO definition all three of these forms fall within the definition of TK, so some confusion can be expected here. But what some states reported as legislation regarding TK actually turned out to be legislation regarding Genetic Resources 82. Some legislation turned to provide a protection beyond what is defined as TK, for instance also certain human remains 83 or even living individuals 84 were protected under reported legislation. Also legislation regarding geographical indications of origin 85 was reported. In defence of these states, these particular legislations were reported together with other legislations that did regard actual TK, so they could be assumed to be reported for thoroughness sake more than anything else. This does show, however, that even though WIPO has provided an elaborate working definition for TK, there still is quite unclarity about what really is and, maybe even more important, is not TK Regional preferences Another point that stands out before any statistical analysis is done on the database, is the difference in focus of the legislation. A lot of legislation regarding Biological TK was reported, relative to other forms of TK. Just to give a first impression of this, Figure 4 shows a histogram. This histogram shows the relative number of states per region with 82 For example Switzerland, Portugal and Bhutan. 83 In the Aboriginal and Torres Strait Islander Protection Act, 1984, Australia, and the Native American Graves Protection and Repatriation Act (NAGRPA), 1990, USA. 84 In the Historical and Cultural Preservation Act, 1995, Palau. 85 Decreto Legislativo N 1075: Decreto Legislativo que aprueba Disposiciones Complementarias a la Decisión 486 de la Comisión de la Comunidad Andina que establece el Régimen Común sobre Propiedad Industrial, Peru. 52

69 legislation regarding Biological TK on the one hand and legislation regarding other forms of TK on the other hand 86. Figure 4. Relative number of states per region that have reported legislation biological TK TK What this shows is that a relative large number of South-American states 87 have legislation protecting Biological TK 88, while in Oceania the focus is much more towards non-biological TK 89, as there are significantly more states in Oceania with such legislation than in other regions 90. Europe has significantly less legislation on TK protection 91. Explanations for these relations could be sought in the presence of specific forms of TK in certain regions: there is probably more non-biological TK in Oceania than in Europe, and there is a lot of biological TK to be found in South-America, and hence more biological TK to be protected. But this can by no means be the full explanation of these variations, as at first sight one might suspect that the presence of TK and biological TK in regions such 86 The correlation between the regions and the reported legislation of TK and biological TK are listed in appendix % of all South-American states. 88 The correlation coefficient between the variable South America and the reporting of biological TK is significant at a level of 2% % of the states in Oceania have legislation protecting non-biological TK. 90 With a significance level of 1% the correlation coefficient between the variable Oceania and the reporting of legislation on TK is significantly different from % of all European states have reported legislation on non-biological TK, the correlation coefficient between the variable Europe and TK is significant at a level of 1%. 53